ICML
Collection
Accepted papers for ICML (International Conference on Machine Learning), one dataset per year. • 14 items • Updated
paper_id stringlengths 10 10 | title stringlengths 15 163 | paper_url stringlengths 42 42 | authors listlengths 1 40 | type stringclasses 3
values | primary_area stringclasses 84
values | abstract large_stringlengths 393 2.6k | keywords listlengths 1 20 | TL;DR large_stringlengths 7 250 ⌀ | submission_number int64 1 16.4k | arxiv_id stringlengths 10 10 ⌀ | arxiv_id_source stringclasses 3
values |
|---|---|---|---|---|---|---|---|---|---|---|---|
U8GUmxnzXn | UnHiPPO: Uncertainty-aware Initialization for State Space Models | https://openreview.net/forum?id=U8GUmxnzXn | [
"Marten Lienen",
"Abdullah Saydemir",
"Stephan Günnemann"
] | Poster | deep_learning->sequential_models_time_series | State space models are emerging as a dominant model class for sequence problems with many relying on the HiPPO framework to initialize their dynamics. However, HiPPO fundamentally assumes data to be noise-free; an assumption often violated in practice. We extend the HiPPO theory with measurement noise and derive an unc... | [
"state space",
"uncertainty",
"hippo",
"mamba",
"kalman",
"noise",
"filter"
] | HiPPO extension based on linear stochastic control theory and the Kalman filter making SSMs more robust against noise | 16,431 | 2506.05065 | title_snapshot |
Dqp6IMI3gQ | When Will It Fail?: Anomaly to Prompt for Forecasting Future Anomalies in Time Series | https://openreview.net/forum?id=Dqp6IMI3gQ | [
"Min-Yeong Park",
"Won-Jeong Lee",
"Seong Tae Kim",
"Gyeong-Moon Park"
] | Poster | deep_learning->sequential_models_time_series | Recently, forecasting future abnormal events has emerged as an important scenario to tackle realworld necessities. However, the solution of predicting specific future time points when anomalies will occur, known as Anomaly Prediction (AP), remains under-explored. Existing methods dealing with time series data fail in A... | [
"Time series forecasting",
"time series anomaly detection"
] | A2P: See anomalies before they strike! | 16,430 | 2506.23596 | title_snapshot |
GKZySvM2t9 | KGMark: A Diffusion Watermark for Knowledge Graphs | https://openreview.net/forum?id=GKZySvM2t9 | [
"Hongrui Peng",
"Haolang Lu",
"Yuanlong Yu",
"WeiYe Fu",
"Kun Wang",
"Guoshun Nan"
] | Poster | social_aspects->fairness | Knowledge graphs (KGs) are ubiquitous in numerous real-world applications, and watermarking facilitates protecting intellectual property and preventing potential harm from AI-generated content. Existing watermarking methods mainly focus on static plain text or image data, while they can hardly be applied to dynamic gra... | [
"Watermarking",
"Knowledge Graph",
"Diffusion Models",
"Generative Models"
] | We present KGMark, the first watermarking method for knowledge graph embeddings that ensures high detectability, transparency, and robustness across various graph modifications. | 16,409 | 2505.23873 | title_snapshot |
GdYg0Ohx0k | LSCD: Lomb--Scargle Conditioned Diffusion for Time series Imputation | https://openreview.net/forum?id=GdYg0Ohx0k | [
"Elizabeth Fons",
"Alejandro Sztrajman",
"Yousef El-Laham",
"Luciana Ferrer",
"Svitlana Vyetrenko",
"Manuela Veloso"
] | Poster | deep_learning->sequential_models_time_series | Time series with missing or irregularly sampled data are a persistent challenge in machine learning. Many methods operate on the frequency-domain, relying on the Fast Fourier Transform (FFT) which assumes uniform sampling, therefore requiring prior interpolation that can distort the spectra. To address this limitation,... | [
"time series",
"diffusion models",
"frequency spectrum"
] | We propose Lomb–Scargle Conditioned Diffusion (LSCD), a diffusion-based time series imputation method that leverages a differentiable Lomb–Scargle periodogram to handle irregular sampling and preserve spectral consistency | 16,408 | 2506.17039 | title_snapshot |
Q0rKYiVEZq | Emoji Attack: Enhancing Jailbreak Attacks Against Judge LLM Detection | https://openreview.net/forum?id=Q0rKYiVEZq | [
"Zhipeng Wei",
"Yuqi Liu",
"N. Benjamin Erichson"
] | Poster | social_aspects->safety | Jailbreaking techniques trick Large Language Models (LLMs) into producing restricted output, posing a potential threat. One line of defense is to use another LLM as a Judge to evaluate the harmfulness of generated text. However, we reveal that these Judge LLMs are vulnerable to token segmentation bias, an issue that ar... | [
"LLM safety; Jailbreaking Attacks; Judge LLMs; Token Segmentation"
] | We introduce Emoji Attack, an adversarial strategy that exploits token segmentation bias in Judge LLMs by inserting emojis to manipulate tokenization, enhancing the effectiveness of jailbreak attacks against Judge LLM detection. | 16,371 | 2411.01077 | title_snapshot |
mEV0nvHcK3 | Towards Practical Defect-Focused Automated Code Review | https://openreview.net/forum?id=mEV0nvHcK3 | [
"Junyi Lu",
"Lili Jiang",
"Xiaojia Li",
"Jianbing Fang",
"Fengjun Zhang",
"Li Yang",
"Chun Zuo"
] | Spotlight | applications | The complexity of code reviews has driven efforts to automate review comments, but prior approaches oversimplify this task by treating it as snippet-level code-to-text generation and relying on text similarity metrics like BLEU for evaluation. These methods overlook repository context, real-world merge request evaluati... | [
"Automated Code Review",
"Merge Request Analysis",
"Large Language Models (LLMs)",
"Defect Detection",
"Evaluation Metrics for Code Review",
"Code Context Extraction",
"Multi-Agent LLM Collaboration"
] | This work presents an end-to-end approach to automated code review that goes beyond snippet-level generation and text-similarity metrics, achieving significant gains over existing baselines in real-world, industry-scale codebases. | 16,368 | 2505.17928 | title_snapshot |
rnx11J4hsg | HiRemate: Hierarchical Approach for Efficient Re-materialization of Neural Networks | https://openreview.net/forum?id=rnx11J4hsg | [
"Julia Gusak",
"Xunyi Zhao",
"Théotime Le Hellard",
"Zhe LI",
"Lionel Eyraud-Dubois",
"Olivier Beaumont"
] | Poster | general_machine_learning->hardware_and_software | Training deep neural networks (DNNs) on memory-limited GPUs is challenging, as storing intermediate activations often exceeds available memory. Re-materialization, a technique that preserves exact computations, addresses this by selectively recomputing activations instead of storing them. However, existing methods eit... | [
"Rematerialization",
"Checkpointing",
"Memory-Efficient Training",
"Neural Networks",
"PyTorch",
"Integer Linear Programming",
"Training"
] | null | 16,364 | null | null |
GByP03IitA | ITFormer: Bridging Time Series and Natural Language for Multi-Modal QA with Large-Scale Multitask Dataset | https://openreview.net/forum?id=GByP03IitA | [
"Yilin wang",
"Peixuan Lei",
"Jie Song",
"Yuzhe Hao",
"Tao Chen",
"Yuxuan Zhang",
"LEI JIA",
"Yuanxiang Li",
"zhongyu wei"
] | Poster | applications->time_series | Time-series data are critical in diverse applications, such as industrial monitoring, medical diagnostics, and climate research. However, effectively integrating these high-dimensional temporal signals with natural language for dynamic, interactive tasks remains a significant challenge. To address this, we introduce th... | [
"Time Series Analysis",
"Time-Series Question Answering",
"Time-Series-Textual Alignment",
"Time-Series-Textual Fusion"
] | Bridging time-series data and natural language, we propose ITFormer and introduce EngineMT-QA, enabling efficient and accurate Time-Series Question Answering for multimodal AI | 16,325 | 2506.20093 | title_snapshot |
7QFmZ7i7sr | GPEN: Global Position Encoding Network for Enhanced Subgraph Representation Learning | https://openreview.net/forum?id=7QFmZ7i7sr | [
"Nannan Wu",
"Yuming Huang",
"Yiming Zhao",
"Jie Chen",
"Wenjun Wang"
] | Poster | deep_learning->graph_neural_networks | Subgraph representation learning has attracted growing interest due to its wide applications in various domains. However, existing methods primarily focus on local neighborhood structures while overlooking the significant impact of global structural information, in particular the influence of multi-hop neighbors beyond... | [
"Subgraph Representation Learning"
] | null | 16,318 | null | null |
w2QNIkcwWw | Fast Min-$\epsilon$ Segmented Regression using Constant-Time Segment Merging | https://openreview.net/forum?id=w2QNIkcwWw | [
"Ansgar Lößer",
"Max Schlecht",
"Florian Schintke",
"Joel Witzke",
"Matthias Weidlich",
"Björn Scheuermann"
] | Poster | general_machine_learning | Segmented regression is a statistical method that approximates a function $f$ by a piecewise function $\hat{f}$ using noisy data samples.
*Min-$\epsilon$* approaches aim to reduce the regression function's mean squared error (MSE) for a given number of $k$ segments.
An optimal solution for *min-$\epsilon$* segmented re... | [
"Regression",
"Segmented Regression",
"Time-Series Analysis"
] | null | 16,316 | null | null |
1OUEnfusEd | How Compositional Generalization and Creativity Improve as Diffusion Models are Trained | https://openreview.net/forum?id=1OUEnfusEd | [
"Alessandro Favero",
"Antonio Sclocchi",
"Francesco Cagnetta",
"Pascal Frossard",
"Matthieu Wyart"
] | Poster | deep_learning | Natural data is often organized as a hierarchical composition of features. How many samples do generative models need in order to learn the composition rules, so as to produce a combinatorially large number of novel data? What signal in the data is exploited to learn those rules? We investigate these questions in the c... | [
"Science of deep learning",
"compositionality",
"diffusion models",
"probabilistic graphical models",
"sample complexity",
"generalization"
] | null | 16,314 | 2502.12089 | title_snapshot |
P0wSGDoip1 | Gradient-based Explanations for Deep Learning Survival Models | https://openreview.net/forum?id=P0wSGDoip1 | [
"Sophie Hanna Langbein",
"Niklas Koenen",
"Marvin N. Wright"
] | Poster | applications->health_medicine | Deep learning survival models often outperform classical methods in time-to-event predictions, particularly in personalized medicine, but their "black box" nature hinders broader adoption. We propose a framework for gradient-based explanation methods tailored to survival neural networks, extending their use beyond regr... | [
"Deep Learning",
"Survival Analysis",
"Explainable Artificial Intelligence",
"Interpretable Machine Learning",
"XAI",
"IML",
"Feature Attribution"
] | We introduce gradient-based explanation methods for survival neural networks, offering improved interpretability, time-dependent insights, and faster, more accurate alternatives for analyzing medical and multi-modal data. | 16,311 | 2502.04970 | title_snapshot |
9Kywz2fO26 | Pairwise Maximum Likelihood For Multi-Class Logistic Regression Model With Multiple Rare Classes | https://openreview.net/forum?id=9Kywz2fO26 | [
"Xuetong Li",
"Danyang Huang",
"Hansheng Wang"
] | Poster | applications->everything_else | We study in this work the problem of multi-class logistic regression with one major class and multiple rare classes, which is motivated by a real application in TikTok live stream data. The model is inspired by the two-class logistic regression model of Wang (2020) but with surprising theoretical findings, which in tur... | [
"Multi-class logistic regression model",
"Pairwise maximum likelihood estimation",
"Rare class analysis",
"Car plate recognition"
] | We study multi-class logistic regression with multiple rare classes, proposing an efficient and effective parallel pairwise estimation method. | 16,293 | null | null |
DkRYImuQA9 | ShieldAgent: Shielding Agents via Verifiable Safety Policy Reasoning | https://openreview.net/forum?id=DkRYImuQA9 | [
"Zhaorun Chen",
"Mintong Kang",
"Bo Li"
] | Poster | social_aspects->safety | Autonomous agents powered by foundation models have seen widespread adoption across various real-world applications. However, they remain highly vulnerable to malicious instructions and attacks, which can result in severe consequences such as privacy breaches and financial losses. More critically, existing guardrails f... | [
"LLM Agent Safety",
"LLM Guardrail Agent",
"Policy Compliance",
"Automated Logic Reasoning"
] | We present ShieldAgent, which safeguards foundation model agents by enforcing policy compliance, and ShieldAgent-Bench, a dataset for evaluating guardrail performance across diverse real-world scenarios. | 16,287 | 2503.22738 | title_snapshot |
Uc0dTE2Wox | Rethinking Benign Overfitting in Two-Layer Neural Networks | https://openreview.net/forum?id=Uc0dTE2Wox | [
"Ruichen Xu",
"Kexin Chen"
] | Poster | deep_learning->theory | Recent theoretical studies (Kou et al., 2023; Cao et al., 2022) revealed a sharp phase transition from benign to harmful overfitting when the
noise-to-feature ratio exceeds a threshold—a situation common in long-tailed data distributions where atypical data is prevalent. However, such harmful overfitting rarely happens... | [
"Benign overfitting",
"long-tailed data",
"two-layer neural networks"
] | We re-examine benign overfitting in two-layer neural networks and prove that while data can be classified by explicit features, long-tailed data can also be classified base on implicit features learned from class-dependent noise. | 16,273 | 2502.11893 | title_snapshot |
JD4eHocSPi | Symmetry-Aware GFlowNets | https://openreview.net/forum?id=JD4eHocSPi | [
"Hohyun Kim",
"Seunggeun Lee",
"Min-hwan Oh"
] | Poster | deep_learning->generative_models_and_autoencoders | Generative Flow Networks (GFlowNets) offer a powerful framework for sampling graphs in proportion to their rewards. However, existing approaches suffer from systematic biases due to inaccuracies in state transition probability computations. These biases, rooted in the inherent symmetries of graphs, impact both atom-bas... | [
"GFlowNet",
"graph generation",
"molecule optimization"
] | This paper analyzes the bias inherent in GFlowNets and proposes a reward-scaling method to address the issue. | 16,266 | 2506.02685 | title_snapshot |
7J1kwZY72h | Kinetic Langevin Diffusion for Crystalline Materials Generation | https://openreview.net/forum?id=7J1kwZY72h | [
"François R J Cornet",
"Federico Bergamin",
"Arghya Bhowmik",
"Juan Maria Garcia-Lastra",
"Jes Frellsen",
"Mikkel N. Schmidt"
] | Poster | deep_learning->generative_models_and_autoencoders | Generative modeling of crystalline materials using diffusion models presents a series of challenges: the data distribution is characterized by inherent symmetries and involves multiple modalities, with some defined on specific manifolds. Notably, the treatment of fractional coordinates representing atomic positions in ... | [
"generative models",
"diffusion models",
"crystals",
"materials"
] | null | 16,254 | 2507.03602 | title_snapshot |
FSlTEObdLl | CombiMOTS: Combinatorial Multi-Objective Tree Search for Dual-Target Molecule Generation | https://openreview.net/forum?id=FSlTEObdLl | [
"Thibaud Southiratn",
"Bonil Koo",
"Yijingxiu Lu",
"Sun Kim"
] | Poster | applications->health_medicine | Dual-target molecule generation, which focuses on discovering compounds capable of interacting with two target proteins, has garnered significant attention due to its potential for improving therapeutic efficiency, safety and resistance mitigation.
Existing approaches face two critical challenges.
First, by simplifying... | [
"Dual-target Molecule Generation",
"Fragment-based Drug Discovery",
"Monte-Carlo Tree Search",
"Pareto Optimization",
"Search Space Reduction"
] | CombiMOTS is a framework that designs dual-target compounds using industrial building blocks and Pareto MCTS for favorable multi-objective trade-offs and improved synthesizability. | 16,227 | 2604.23307 | title_snapshot |
NWKjVzkDzg | Scalable Meta-Learning via Mixed-Mode Differentiation | https://openreview.net/forum?id=NWKjVzkDzg | [
"Iurii Kemaev",
"Dan A. Calian",
"Luisa M Zintgraf",
"Gregory Farquhar",
"Hado van Hasselt"
] | Poster | deep_learning->algorithms | Gradient-based bilevel optimisation is a powerful technique with applications in hyperparameter optimisation, task adaptation, algorithm discovery, meta-learning more broadly, and beyond. It often requires differentiating through the gradient-based optimisation process itself, leading to "gradient-of-a-gradient" calcul... | [
"meta-learning",
"bilevel optimization",
"second-order optimization",
"automatic differentiation",
"autodiff",
"mixed-mode differentiation",
"gradient-based methods",
"scalable algorithms"
] | A practical algorithm that uses mixed-mode differentiation to construct more efficient and scalable computational graphs yielding over 10x memory and up to 25% wall-clock time improvements over standard implementations in modern meta-learning setups. | 16,225 | 2505.00793 | title_snapshot |
RO5OGOzs6M | PINNsAgent: Automated PDE Surrogation with Large Language Models | https://openreview.net/forum?id=RO5OGOzs6M | [
"Qingpo Wuwu",
"Chonghan Gao",
"Tianyu Chen",
"Yihang Huang",
"Yuekai Zhang",
"Jianing Wang",
"Jianxin Li",
"Haoyi Zhou",
"Shanghang Zhang"
] | Poster | applications->chemistry_physics_and_earth_sciences | Solving partial differential equations (PDEs) using neural methods has been a long-standing scientific and engineering research pursuit. Physics-Informed Neural Networks (PINNs) have emerged as a promising alternative to traditional numerical methods for solving PDEs. However, the gap between domain-specific knowledge ... | [
"pinns",
"llm-agent"
] | null | 16,222 | 2501.12053 | title_snapshot |
rkHCHI5H5W | Compositional Generalization via Forced Rendering of Disentangled Latents | https://openreview.net/forum?id=rkHCHI5H5W | [
"Qiyao Liang",
"Daoyuan Qian",
"Liu Ziyin",
"Ila R Fiete"
] | Poster | deep_learning->everything_else | Composition—the ability to generate myriad variations from finite means—is believed to underlie powerful generalization. However, compositional generalization remains a key challenge for deep learning. A widely held assumption is that learning disentangled (factorized) representations naturally supports this kind of ex... | [
"Factorization",
"Compositionality",
"Compositional Generalization",
"Data Efficiency"
] | Standard networks given separate x/y cues still memorize and stitch examples rather than combine them. Forcing each cue into the output space—via low-rank embeddings or simple stripe data—enables true compositional generalization. | 16,212 | 2501.18797 | title_snapshot |
il3KRr4H9u | BaxBench: Can LLMs Generate Correct and Secure Backends? | https://openreview.net/forum?id=il3KRr4H9u | [
"Mark Vero",
"Niels Mündler",
"Victor Chibotaru",
"Veselin Raychev",
"Maximilian Baader",
"Nikola Jovanović",
"Jingxuan He",
"Martin Vechev"
] | Spotlight | social_aspects->security | Automatic program generation has long been a fundamental challenge in computer science. Recent benchmarks have shown that large language models (LLMs) can effectively generate code at the function level, make code edits, and solve algorithmic coding tasks. However, to achieve full automation, LLMs should be able to gen... | [
"large language model",
"large language models",
"LLM",
"code generation",
"code security",
"security",
"benchmark"
] | This paper introduces a novel benchmark to measure the correctness and security of LLM-generated code for backend applications. | 16,211 | 2502.11844 | title_snapshot |
aqZKgwf7Cc | Time to Spike? Understanding the Representational Power of Spiking Neural Networks in Discrete Time | https://openreview.net/forum?id=aqZKgwf7Cc | [
"Duc Anh Nguyen",
"Ernesto Araya",
"Adalbert Fono",
"Gitta Kutyniok"
] | Poster | theory->deep_learning | Recent years have seen significant progress in developing spiking neural networks (SNNs) as a potential solution to the energy challenges posed by conventional artificial neural networks (ANNs). However, our theoretical understanding of SNNs remains relatively limited compared to the ever-growing body of literature on ... | [
"Spiking neural networks",
"Linear regions",
"Approximation theory"
] | null | 16,203 | 2505.18023 | title_snapshot |
l5KpQ5MmaD | Scalable Non-Equivariant 3D Molecule Generation via Rotational Alignment | https://openreview.net/forum?id=l5KpQ5MmaD | [
"Yuhui Ding",
"Thomas Hofmann"
] | Poster | deep_learning->generative_models_and_autoencoders | Equivariant diffusion models have achieved impressive performance in 3D molecule generation. These models incorporate Euclidean symmetries of 3D molecules by utilizing an SE(3)-equivariant denoising network. However, specialized equivariant architectures limit the scalability and efficiency of diffusion models. In this... | [
"Non-equivariant diffusion",
"3D molecule generation"
] | null | 16,184 | 2506.10186 | title_snapshot |
SOwcmZ91Sl | Learning Distances from Data with Normalizing Flows and Score Matching | https://openreview.net/forum?id=SOwcmZ91Sl | [
"Peter Sorrenson",
"Daniel Behrend-Uriarte",
"Christoph Schnoerr",
"Ullrich Koethe"
] | Poster | general_machine_learning->representation_learning | Density-based distances (DBDs) provide a principled approach to metric learning by defining distances in terms of the underlying data distribution. By employing a Riemannian metric that increases in regions of low probability density, shortest paths naturally follow the data manifold. Fermat distances, a specific type ... | [
"density-based distance",
"Fermat distance",
"Riemannian geometry",
"representation learning",
"normalizing flows",
"score matching"
] | Our work improves estimation of Fermat distances by combining normalizing flows, score-based models, geodesic smoothing, and a new dimension-adapted Fermat distance for better scalability. | 16,182 | 2407.09297 | title_snapshot |
edhBkkYS8R | The Importance of Being Lazy: Scaling Limits of Continual Learning | https://openreview.net/forum?id=edhBkkYS8R | [
"Jacopo Graldi",
"Alessandro Breccia",
"Giulia Lanzillotta",
"Thomas Hofmann",
"Lorenzo Noci"
] | Poster | deep_learning | Despite recent efforts, neural networks still struggle to learn in non-stationary environments, and our understanding of catastrophic forgetting (CF) is far from complete.
In this work, we perform a systematic study on the impact of model scale and the degree of feature learning in continual learning. We reconcile exis... | [
"feature learning",
"continual learning",
"deep learning",
"forgetting",
"ntk",
"muP",
"lazy",
"dmft"
] | We investigate how network size and feature learning regimes affect catastrophic forgetting in continual learning. | 16,178 | 2506.16884 | title_snapshot |
YUtJsxQjv3 | EcoMapper: Generative Modeling for Climate-Aware Satellite Imagery | https://openreview.net/forum?id=YUtJsxQjv3 | [
"Muhammed Goktepe",
"Amir hossein Shamseddin",
"Erencan Uysal",
"Javier Muinelo Monteagudo",
"Lukas Drees",
"Aysim Toker",
"Senthold Asseng",
"Malte von Bloh"
] | Poster | applications->chemistry_physics_and_earth_sciences | Satellite imagery is essential for Earth observation, enabling applications like crop yield prediction, environmental monitoring, and climate
change assessment. However, integrating satellite imagery with climate data remains a challenge, limiting its utility for forecasting and scenario analysis. We introduce a novel ... | [
"remote sensing",
"generative modeling",
"stable diffusion",
"climate",
"satellite imagery",
"computer vision"
] | We present a novel dataset and two generative approaches using fine-tuned Stable Diffusion 3 models to create realistic satellite imagery conditioned on climate and land cover data, enabling single-image and conditional time-series generation. | 16,158 | null | null |
umT6rMf1Rm | DEFAME: Dynamic Evidence-based FAct-checking with Multimodal Experts | https://openreview.net/forum?id=umT6rMf1Rm | [
"Tobias Braun",
"Mark Rothermel",
"Marcus Rohrbach",
"Anna Rohrbach"
] | Poster | applications->everything_else | The proliferation of disinformation demands reliable and scalable fact-checking solutions. We present **D**ynamic **E**vidence-based **FA**ct-checking with **M**ultimodal **E**xperts (DEFAME), a modular, zero-shot MLLM pipeline for open-domain, text-image claim verification. DEFAME operates in a six-stage process, dyna... | [
"fact-checking",
"multimodal",
"claim-verification",
"MLLM"
] | We introduce DEFAME, a modular, training-free fact-checking system that verifies open-domain text-image claims end-to-end, dynamically retrieving and integrating textual and visual evidence. | 16,152 | 2412.10510 | title_snapshot |
Au9rfI6Fjd | Generalization of noisy SGD in unbounded non-convex settings | https://openreview.net/forum?id=Au9rfI6Fjd | [
"Leello Tadesse Dadi",
"Volkan Cevher"
] | Poster | theory->optimization | We study generalization of iterative noisy gradient schemes on smooth non-convex losses. Formally, we establish time-independent information theoretic generalization bounds for Stochastic Gradient Langevin Dynamics (SGLD) that do not diverge as the iteration count increases. Our bounds are obtained through a stability ... | [
"Information theoretic generalization",
"Langevin",
"SGD",
"differential privacy"
] | null | 16,148 | null | null |
ECayXPDoha | Statistical Hypothesis Testing for Auditing Robustness in Language Models | https://openreview.net/forum?id=ECayXPDoha | [
"Paulius Rauba",
"Qiyao Wei",
"Mihaela van der Schaar"
] | Poster | general_machine_learning->everything_else | Consider the problem of testing whether the outputs of a large language model (LLM) system change under an arbitrary intervention, such as an input perturbation or changing the model variant. We cannot simply compare two LLM outputs since they might differ due to the stochastic nature of the system, nor can we compare ... | [
"language models",
"safety",
"interpretability",
"reliability"
] | We develop a statitical hypothesis testing method to quantify the impact of a perturbation in the input prompt on the outputs of language models | 16,143 | 2506.07947 | title_snapshot |
0Hd1lh52Fi | SDE Matching: Scalable and Simulation-Free Training of Latent Stochastic Differential Equations | https://openreview.net/forum?id=0Hd1lh52Fi | [
"Grigory Bartosh",
"Dmitry Vetrov",
"Christian A. Naesseth"
] | Poster | deep_learning->sequential_models_time_series | The Latent Stochastic Differential Equation (SDE) is a powerful tool for time series and sequence modeling. However, training Latent SDEs typically relies on adjoint sensitivity methods, which depend on simulation and backpropagation through approximate SDE solutions, which limit scalability. In this work, we propose S... | [
"diffusion",
"generative models",
"SDE",
"time series",
"variational inference"
] | null | 16,133 | 2502.02472 | title_snapshot |
DTdtM53iag | An Empirical Study on Configuring In-Context Learning Demonstrations for Unleashing MLLMs' Sentimental Perception Capability | https://openreview.net/forum?id=DTdtM53iag | [
"Daiqing Wu",
"Dongbao Yang",
"Sicheng Zhao",
"Can Ma",
"Yu Zhou"
] | Poster | applications->computer_vision | The advancements in Multimodal Large Language Models (MLLMs) have enabled various multimodal tasks to be addressed under a zero-shot paradigm. This paradigm sidesteps the cost of model fine-tuning, emerging as a dominant trend in practical application. Nevertheless, Multimodal Sentiment Analysis (MSA), a pivotal challe... | [
"Multimodal Sentiment Analysis",
"Multimodal Large Language Model",
"In-Context Learning"
] | We conduct an in-depth investigation into three pivotal factors that influence the configuration of In-Context Learning demonstrations on Multimodal Sentiment Analysis. | 16,131 | 2505.16193 | title_snapshot |
G80YGyxzv7 | Beyond Cropped Regions: New Benchmark and Corresponding Baseline for Chinese Scene Text Retrieval in Diverse Layouts | https://openreview.net/forum?id=G80YGyxzv7 | [
"Gengluo Li",
"Huawen Shen",
"Yu Zhou"
] | Poster | applications->computer_vision | Chinese scene text retrieval is a practical task that aims to search for images containing visual instances of a Chinese query text. This task is extremely challenging because Chinese text often features complex and diverse layouts in real-world scenes. Current efforts tend to inherit the solution for English scene tex... | [
"Scene text retrieval",
"Multimodal retrieval",
"Text understanding",
"Text layout"
] | We introduce a new benchmark for Chinese scene text retrieval, highlighting the limitations of previous methods and proposing an approach that outperforms existing techniques. | 16,125 | 2506.04999 | title_snapshot |
cEKrGCFXPA | Controlling Large Language Model with Latent Action | https://openreview.net/forum?id=cEKrGCFXPA | [
"Chengxing Jia",
"Ziniu Li",
"Pengyuan Wang",
"Yi-Chen Li",
"Zhenyu Hou",
"Yuxiao Dong",
"Yang Yu"
] | Poster | deep_learning->large_language_models | Adapting Large Language Models (LLMs) to downstream tasks using Reinforcement Learning (RL) has proven to be an effective approach. However, LLMs do not inherently define the structure of an agent for RL training, particularly in terms of specifying the action space. This paper studies learning a compact latent action ... | [
"Controllable language model",
"latent action model"
] | Develop a new architecture of language model to control language with latent action | 16,114 | 2503.21383 | manual |
gmdElnwBxt | Neutral residues: revisiting adapters for model extension | https://openreview.net/forum?id=gmdElnwBxt | [
"Franck SIGNE TALLA",
"Edouard Grave",
"Herve Jegou"
] | Poster | deep_learning->large_language_models | We address the problem of extending a pre-trained large language model to a new domain that was not seen during training. Standard techniques, such as fine-tuning or low-rank adaptation (LoRA) are successful at domain adaptation, but do not formally add capacity to the model. This often leads to a trade-off, between pe... | [
"LLM",
"Adapters",
"Domain Adaptation",
"Catastrophic Forgetting"
] | We add new capabilities to a pre-trained large language model, without forgetting its original knowledge. | 16,101 | 2410.02744 | title_snapshot |
viXwXCkA7N | Certification for Differentially Private Prediction in Gradient-Based Training | https://openreview.net/forum?id=viXwXCkA7N | [
"Matthew Robert Wicker",
"Philip Sosnin",
"Igor Shilov",
"Adrianna Janik",
"Mark Niklas Mueller",
"Yves-Alexandre de Montjoye",
"Adrian Weller",
"Calvin Tsay"
] | Poster | social_aspects->privacy | We study private prediction where differential privacy is achieved by adding noise to the outputs of a non-private model. Existing methods rely on noise proportional to the global sensitivity of the model, often resulting in sub-optimal privacy-utility trade-offs compared to private training. We introduce a novel appro... | [
"Differential Privacy",
"Convex Optimization",
"Deep Learning"
] | We utilize novel bound propagation algorithms to upper-bound local sensitivity of machine learning predictions and leverage this bound for improved private prediction. | 16,097 | 2406.13433 | title_snapshot |
X21P8etjWL | Measuring In-Context Computation Complexity via Hidden State Prediction | https://openreview.net/forum?id=X21P8etjWL | [
"Vincent Herrmann",
"Róbert Csordás",
"Jürgen Schmidhuber"
] | Poster | deep_learning->sequential_models_time_series | Detecting when a neural sequence model does "interesting" computation is an open problem. The next token prediction loss is a poor indicator: Low loss can stem from trivially predictable sequences that are uninteresting, while high loss may reflect unpredictable but also irrelevant information that can be ignored by th... | [
"in-context learning",
"interpretability",
"transformers"
] | Hidden state unpredictability in sequence models is a meaningful measure for in-context reasoning complexity | 16,096 | 2503.13431 | title_snapshot |
zFR5aWGaUs | Let LLM Tell What to Prune and How Much to Prune | https://openreview.net/forum?id=zFR5aWGaUs | [
"Mingzhe Yang",
"Sihao Lin",
"Changlin Li",
"Xiaojun Chang"
] | Poster | deep_learning->large_language_models | Large language models (LLMs) have revolutionized various AI applications. However, their billions of parameters pose significant challenges for practical deployment. Structured pruning is a hardware-friendly compression technique and receives widespread attention. Nonetheless, existing literature typically targets a si... | [
"large language models",
"model compression",
"network pruning",
"structured pruning"
] | We propose a pruning method that targets multiple LLM modules with dynamic pruning ratios by quantifying the complex interactions within LLMs, achieving better trade-off between efficiency and performance. | 16,094 | null | null |
Dd7Qo7TJpf | Multilayer Matrix Factorization via Dimension-Reducing Diffusion Variational Inference | https://openreview.net/forum?id=Dd7Qo7TJpf | [
"Junbin Liu",
"Farzan Farnia",
"Wing-Kin Ma"
] | Poster | probabilistic_methods->variational_inference | Multilayer matrix factorization (MMF) has recently emerged as a generalized model of, and potentially a more expressive approach than, the classic matrix factorization.
This paper considers MMF under a probabilistic formulation, and our focus is on inference methods under variational inference.
The challenge in this co... | [
"Multilayer matrix factorization",
"Variational inference",
"Variational diffusion models",
"Dimension reduction"
] | null | 16,092 | null | null |
9Klg7ce8D7 | Compressing tree ensembles through Level-wise Optimization and Pruning | https://openreview.net/forum?id=9Klg7ce8D7 | [
"Laurens Devos",
"Timo Martens",
"Deniz Can Oruc",
"Wannes Meert",
"Hendrik Blockeel",
"Jesse Davis"
] | Poster | general_machine_learning->supervised_learning | Tree ensembles (e.g., gradient boosting decision trees) are often used in practice because they offer excellent predictive performance while still being easy and efficient to learn. In some contexts, it is important to additionally optimize their size: this is specifically the case when models need to have verifiable p... | [
"ensembles",
"decision forests",
"model efficiency",
"energy efficiency",
"verification",
"compression"
] | A method is proposed that can reduce the size of decision tree ensembles by orders of magnitude with negligible cost in accuracy. | 16,090 | null | null |
SaKPKyjDp6 | Time Series Representations with Hard-Coded Invariances | https://openreview.net/forum?id=SaKPKyjDp6 | [
"Thibaut Germain",
"Chrysoula Kosma",
"Laurent Oudre"
] | Poster | general_machine_learning->sequential_network_and_time_series_modeling | Automatically extracting robust representations from large and complex time series data is becoming imperative for several real-world applications. Unfortunately, the potential of common neural network architectures in capturing invariant properties of time series remains relatively underexplored. For instance, convolu... | [
"Time Series",
"Invariances",
"Neural Networks",
"Convolutions"
] | Mathematical formalism of time series invariances and design of hard-coded invariant convolutions to commonly observed deformations. | 16,086 | null | null |
2GmXJnyNM4 | Implicit Regularization for Tubal Tensor Factorizations via Gradient Descent | https://openreview.net/forum?id=2GmXJnyNM4 | [
"Santhosh Karnik",
"Anna Veselovska",
"Mark Iwen",
"Felix Krahmer"
] | Oral | theory->learning_theory | We provide a rigorous analysis of implicit regularization in an overparametrized tensor factorization problem beyond the lazy training regime. For matrix factorization problems, this phenomenon has been studied in a number of works. A particular challenge has been to design universal initialization strategies which pro... | [
"overparameterization",
"implicit regularization",
"tensor factorization"
] | We provide a rigorous analysis of implicit regularization in an overparametrized tensor factorization problem beyond the lazy training regime. | 16,047 | 2410.16247 | title_snapshot |
hrBfufwMzg | Voronoi-grid-based Pareto Front Learning and Its Application to Collaborative Federated Learning | https://openreview.net/forum?id=hrBfufwMzg | [
"Mengmeng Chen",
"Xiaohu Wu",
"QIQI LIU",
"Tiantian He",
"Yew-Soon Ong",
"Yaochu Jin",
"Qicheng Lao",
"Han Yu"
] | Poster | applications->everything_else | Multi-objective optimization (MOO) exists extensively in machine learning, and aims to find a set of Pareto-optimal solutions, called the Pareto front, e.g., it is fundamental for multiple avenues of research in federated learning (FL). Pareto-Front Learning (PFL) is a powerful method implemented using Hypernetworks (P... | [
"Hypernetwork",
"Multi-objective optimization",
"Hypervolume",
"Collaborative federated learning"
] | null | 16,040 | 2505.20648 | title_snapshot |
CAPNgWkEEk | Optimal Sensor Scheduling and Selection for Continuous-Discrete Kalman Filtering with Auxiliary Dynamics | https://openreview.net/forum?id=CAPNgWkEEk | [
"Mohamad Al Ahdab",
"John Leth",
"Zheng-Hua Tan"
] | Poster | probabilistic_methods->bayesian_models_and_methods | We study the Continuous-Discrete Kalman Filter (CD-KF) for State-Space Models (SSMs) where continuous-time dynamics are observed via multiple sensors with discrete, irregularly timed measurements. Our focus extends to scenarios in which the measurement process is coupled with the states of an auxiliary SSM. For instanc... | [
"Kalman Filtering",
"Sensor Scheduling",
"Bayesian State-Space Models",
"Control"
] | Scheduling and selecting sensor measurements under varying enviroments in a continuous-discrete kalman filter setup. | 16,017 | 2507.11240 | title_snapshot |
3go0lhfxd0 | Algorithm Development in Neural Networks: Insights from the Streaming Parity Task | https://openreview.net/forum?id=3go0lhfxd0 | [
"Loek van Rossem",
"Andrew M Saxe"
] | Oral | theory->deep_learning | Even when massively overparameterized, deep neural networks show a remarkable ability to generalize. Research on this phenomenon has focused on generalization within distribution, via smooth interpolation. Yet in some settings neural networks also learn to extrapolate to data far beyond the bounds of the original train... | [
"Out-of-distribution generalization",
"Algorithm discovery",
"Deep learning theory",
"Mechanistic Interpretability"
] | We explain in a simple setting how out-of-distribution generalization can occur. | 16,013 | 2507.09897 | title_snapshot |
7jxa1o8rDW | Fairness on Principal Stratum: A New Perspective on Counterfactual Fairness | https://openreview.net/forum?id=7jxa1o8rDW | [
"Haoxuan Li",
"Zeyu Tang",
"Zhichao Jiang",
"Zhuangyan Fang",
"Yue Liu",
"Zhi Geng",
"Kun Zhang"
] | Poster | social_aspects->fairness | Fairness in human and algorithmic decision-making is crucial in areas such as criminal justice, education, and social welfare. Recently, counterfactual fairness has drawn increasing research interest, suggesting that decision-making for individuals should remain the same when intervening with different values on protec... | [
"Counterfactual fairness",
"Principal strata"
] | null | 16,009 | null | null |
kjtvCSkSsy | Exponential Family Variational Flow Matching for Tabular Data Generation | https://openreview.net/forum?id=kjtvCSkSsy | [
"Andrés Guzmán-Cordero",
"Floor Eijkelboom",
"Jan-Willem van de Meent"
] | Poster | deep_learning->generative_models_and_autoencoders | While denoising diffusion and flow matching have driven major advances in generative modeling, their application to tabular data remains limited, despite its ubiquity in real-world applications.
To this end, we develop *TabbyFlow*, a variational Flow Matching (VFM) method for tabular data generation.
To apply VFM to ... | [
"Variational Flow Matching",
"Exponential Families",
"Tabular Data"
] | TabbyFlow extends flow matching to tabular data generation by leveraging exponential family distributions to handle mixed data types efficiently. | 16,008 | 2506.05940 | title_snapshot |
uTv5rOPZr4 | LLMs Can Reason Faster Only If We Let Them | https://openreview.net/forum?id=uTv5rOPZr4 | [
"Bilgehan Sel",
"Lifu Huang",
"Naren Ramakrishnan",
"Ruoxi Jia",
"Ming Jin"
] | Poster | deep_learning->large_language_models | Large language models (LLMs) are making inroads into classical AI problems such as automated planning, yet key shortcomings continue to hamper their integration. Chain-of-Thought (CoT) struggles in complex multi-step reasoning, and Tree-of-Thoughts requires multiple queries that increase computational overhead. Recentl... | [
"large language models",
"decision-making",
"planning"
] | Enabling faster large language model solutions for autonomous reasoning and planning | 15,998 | null | null |
YSVSMV0lXQ | Controlled Generation with Equivariant Variational Flow Matching | https://openreview.net/forum?id=YSVSMV0lXQ | [
"Floor Eijkelboom",
"Heiko Zimmermann",
"Sharvaree Vadgama",
"Erik J Bekkers",
"Max Welling",
"Christian A. Naesseth",
"Jan-Willem van de Meent"
] | Poster | deep_learning->generative_models_and_autoencoders | We derive a controlled generation objective within the framework of Variational Flow Matching (VFM),
which casts flow matching as a variational inference problem.
We demonstrate that controlled generation can be implemented two ways: (1) by way of end-to-end training of conditional generative models, or (2) as a Bayesi... | [
"Variational Flow Matching",
"Conditional Generation",
"Equivariance",
"Molecular Generation"
] | We propose Variational Flow Matching for controlled generation, unifying conditional training and post hoc Bayesian inference with an equivariant formulation for molecular generation. | 15,993 | 2506.18340 | title_snapshot |
hRMAo5N66M | MAGELLAN: Metacognitive predictions of learning progress guide autotelic LLM agents in large goal spaces | https://openreview.net/forum?id=hRMAo5N66M | [
"Loris Gaven",
"Thomas Carta",
"Clément ROMAC",
"Cédric Colas",
"sylvain lamprier",
"Olivier Sigaud",
"Pierre-Yves Oudeyer"
] | Poster | deep_learning->large_language_models | Open-ended learning agents must efficiently prioritize goals in vast possibility spaces, focusing on those that maximize learning progress (LP). When such autotelic exploration is achieved by LLM agents trained with online RL in high-dimensional and evolving goal spaces, a key challenge for LP prediction is modeling on... | [
"LLM agents",
"Open-Ended Learning",
"Learning Progress",
"Goal-conditionned RL",
"Automatic Curriculum Learning"
] | We introduce MAGELLAN, a metacognitive framework that lets LLM agents learn to predict their competence and learning progress online to guide their curriculum in large goal spaces. | 15,991 | 2502.07709 | title_snapshot |
OmQcPgq9RN | The Harder Path: Last Iterate Convergence for Uncoupled Learning in Zero-Sum Games with Bandit Feedback | https://openreview.net/forum?id=OmQcPgq9RN | [
"Côme Fiegel",
"Pierre Menard",
"Tadashi Kozuno",
"Michal Valko",
"Vianney Perchet"
] | Poster | general_machine_learning->online_learning_active_learning_and_bandits | We study the problem of learning in zero-sum matrix games with repeated play and bandit feedback.
Specifically, we focus on developing uncoupled algorithms that guarantee, without communication between players, convergence of the last-iterate to a Nash equilibrium. Although the non-bandit case has been studied extensi... | [
"Game theory",
"Bandit",
"Last-iterate convergence",
"Online learning"
] | We show that learning how to play a game is sometimes harder when using uncoupled algorithms. | 15,967 | 2604.16087 | title_snapshot |
H76PMm7hf2 | Towards Efficient Online Tuning of VLM Agents via Counterfactual Soft Reinforcement Learning | https://openreview.net/forum?id=H76PMm7hf2 | [
"Lang Feng",
"Weihao Tan",
"Zhiyi Lyu",
"Longtao Zheng",
"Haiyang Xu",
"Ming Yan",
"Fei Huang",
"Bo An"
] | Poster | deep_learning->foundation_models | Online fine-tuning vision-language model (VLM) agents with reinforcement learning (RL) has shown promise for equipping agents with multi-step, goal-oriented capabilities in dynamic environments. However, their open-ended textual action space and non-end-to-end nature of action generation present significant challenges ... | [
"vision-language model",
"agent",
"reinforcement learning",
"online fine-tuning",
"counterfactual"
] | null | 15,959 | 2505.03792 | title_snapshot |
h5TXCnnEyy | Towards Attributions of Input Variables in a Coalition | https://openreview.net/forum?id=h5TXCnnEyy | [
"Xinhao Zheng",
"Huiqi Deng",
"Quanshi Zhang"
] | Poster | social_aspects->accountability_transparency_and_interpretability | This paper focuses on the fundamental challenge of partitioning input variables in attribution methods for Explainable AI, particularly in Shapley value-based approaches. Previous methods always compute attributions given a predefined partition but lack theoretical guidance on how to form meaningful variable partitions... | [
"Attribution methods",
"Shapley value"
] | This paper proves the internal mechanism for the conflict of attributions computed under different partitions of input variables. | 15,947 | 2309.13411 | title_snapshot |
4LClOWTAth | Neural Guided Diffusion Bridges | https://openreview.net/forum?id=4LClOWTAth | [
"Gefan Yang",
"Frank van der Meulen",
"Stefan Sommer"
] | Poster | probabilistic_methods->variational_inference | We propose a novel method for simulating conditioned diffusion processes (diffusion bridges) in Euclidean spaces. By training a neural network to approximate bridge dynamics, our approach eliminates the need for computationally intensive Markov Chain Monte Carlo (MCMC) methods or reverse-process modeling. Compared to e... | [
"Diffusion bridge",
"variational approximation",
"change of measure"
] | null | 15,945 | 2502.11909 | title_snapshot |
5QAKPBVdFH | Hide & Seek: Transformer Symmetries Obscure Sharpness & Riemannian Geometry Finds It | https://openreview.net/forum?id=5QAKPBVdFH | [
"Marvin F. da Silva",
"Felix Dangel",
"Sageev Oore"
] | Spotlight | deep_learning->theory | The concept of sharpness has been successfully applied to traditional architectures like MLPs and CNNs to predict their generalization. For transformers, however, recent work reported weak correlation between flatness and generalization. We argue that existing sharpness measures fail for transformers, because they have... | [
"generalization",
"symmetry",
"sharpness",
"flatness",
"riemannian geometry",
"loss landscape"
] | null | 15,939 | 2505.05409 | title_snapshot |
d6CTIPrTTC | ELMO : Efficiency via Low-precision and Peak Memory Optimization in Large Output Spaces | https://openreview.net/forum?id=d6CTIPrTTC | [
"Jinbin Zhang",
"Nasib Ullah",
"Erik Schultheis",
"Rohit Babbar"
] | Poster | general_machine_learning->scalable_algorithms | Large output spaces, also referred to as Extreme multilabel classification (XMC), is a setting that arises, e.g., in large-scale tagging and product-to-product recommendation, and is characterized by the number of labels ranging from hundreds of thousands to millions. This means that the linear classification head, usu... | [
"Extreme Multi-Label Classification",
"Low-Precision Training",
"Peak Memory Optimization",
"FLOAT8 training"
] | We propose a pure low-precision training framework for XMC models using BFLOAT16 and FP8, achieving significant GPU memory savings while competing baselines on most public datasets under low-bitwidth constraints. | 15,923 | 2510.11168 | title_snapshot |
yhgcRwJ9Dn | Hyper-Transforming Latent Diffusion Models | https://openreview.net/forum?id=yhgcRwJ9Dn | [
"Ignacio Peis",
"Batuhan Koyuncu",
"Isabel Valera",
"Jes Frellsen"
] | Poster | deep_learning->generative_models_and_autoencoders | We introduce a novel generative framework for functions by integrating Implicit Neural Representations (INRs) and Transformer-based hypernetworks into latent variable models. Unlike prior approaches that rely on MLP-based hypernetworks with scalability limitations, our method employs a Transformer-based decoder to gene... | [
"Latent Diffusion Models",
"Transformers",
"INRs"
] | We propose a generative framework for INRs that integrates a Transformer-based hypernetwork decoder into latent diffusion models, enabling scalable INR generation and efficient adaptation via hyper-transforming, which fine-tunes only the decoder. | 15,913 | 2504.16580 | title_snapshot |
Ce79P8ULPY | Emergent Response Planning in LLMs | https://openreview.net/forum?id=Ce79P8ULPY | [
"Zhichen Dong",
"Zhanhui Zhou",
"Zhixuan Liu",
"Chao Yang",
"Chaochao Lu"
] | Poster | deep_learning->large_language_models | In this work, we argue that large language models (LLMs), though trained to predict only the next token, exhibit emergent planning behaviors: $\textbf{their hidden representations encode future outputs beyond the next token}$. Through simple probing, we demonstrate that LLM prompt representations encode global attribut... | [
"Large Language Models",
"Emergent Planning",
"Model Probing and Hidden Representations"
] | This paper shows that Large Language Models (LLMs) exhibit emergent response planning, as their internal hidden representations encode predictable, global attributes of their entire future output. | 15,911 | 2502.06258 | title_snapshot |
iXvm0zvspb | Explicit Preference Optimization: No Need for an Implicit Reward Model | https://openreview.net/forum?id=iXvm0zvspb | [
"Xiangkun Hu",
"Lemin Kong",
"Tong He",
"David Wipf"
] | Poster | deep_learning->large_language_models | The generated responses of large language models (LLMs) are often fine-tuned to human preferences through a process called reinforcement learning from human feedback (RLHF). As RLHF relies on a challenging training sequence, whereby a separate reward model is independently learned and then later applied to LLM policy ... | [
"direct preference optimization",
"reinforcement learning from human feedback",
"preference alignment",
"regularized regression"
] | This paper introduces and analyzes an alternative to direct preference optimization that relies on no implicit reward. | 15,908 | 2506.07492 | title_snapshot |
CS4RyQuTig | CaDA: Cross-Problem Routing Solver with Constraint-Aware Dual-Attention | https://openreview.net/forum?id=CS4RyQuTig | [
"Han Li",
"Fei Liu",
"Zhi Zheng",
"Yu Zhang",
"Zhenkun Wang"
] | Poster | optimization->discrete_and_combinatorial_optimization | Vehicle routing problems (VRPs) are significant combinatorial optimization problems (COPs) holding substantial practical importance. Recently, neural combinatorial optimization (NCO), which involves training deep learning models on extensive data to learn vehicle routing heuristics, has emerged as a promising approach ... | [
"Vehicle Routing Problem",
"Multi-Task Learning",
"Task-Specific Prompt",
"Dual Attention Mechanism",
"Cross-Problem Learning"
] | null | 15,901 | 2412.00346 | title_snapshot |
GDvO6viRCF | Uncertainty Estimation for Heterophilic Graphs Through the Lens of Information Theory | https://openreview.net/forum?id=GDvO6viRCF | [
"Dominik Fuchsgruber",
"Tom Wollschläger",
"Johannes Bordne",
"Stephan Günnemann"
] | Poster | deep_learning->graph_neural_networks | While uncertainty estimation for graphs recently gained traction, most methods rely on homophily and deteriorate in heterophilic settings.
We address this by analyzing message passing neural networks from an information-theoretic perspective and developing a suitable analog to data processing inequality to quantify... | [
"Heterophilic Graphs",
"Uncertainty Estimation",
"Information Theory",
"Graph Neural Networks"
] | Information theory based joint density estimatior on the hidden representations to quantify uncertainty for heterophilic graphs. | 15,892 | 2505.22152 | title_snapshot |
3Jr5Al16MS | Near Optimal Best Arm Identification for Clustered Bandits | https://openreview.net/forum?id=3Jr5Al16MS | [
"Yash",
"Avishek Ghosh",
"Nikhil Karamchandani"
] | Poster | general_machine_learning->online_learning_active_learning_and_bandits | This work investigates the problem of best arm identification for multi-agent multi-armed bandits. We consider $N$ agents grouped into $M$ clusters, where each cluster solves a stochastic bandit problem. The mapping between agents and bandits is \textit{a priori} unknown. Each bandit is associated with $K$ arms, and th... | [
"Best Arm Identification",
"Multi-Armed Bandits (MAB)",
"Clustered Bandits",
"Pure Exploration"
] | We address the best arm identification problem for Federated (Distributed) Bandits | 15,890 | 2505.10147 | title_snapshot |
rTPq8VzhmZ | High Probability Bound for Cross-Learning Contextual Bandits with Unknown Context Distributions | https://openreview.net/forum?id=rTPq8VzhmZ | [
"Ruiyuan Huang",
"Zengfeng Huang"
] | Poster | theory->online_learning_and_bandits | Motivated by applications in online bidding and sleeping bandits, we examine the problem of contextual bandits with cross learning, where the learner observes the loss associated with the action across all possible contexts, not just the current round’s context. Our focus is on a setting where losses are chosen adversa... | [
"bandit",
"contextual bandit",
"cross-learning",
"high probability bounds"
] | We give a nearly optimal high probability bound for the cross-learning contextual bandits with unknown context distributions. | 15,881 | 2410.04080 | title_snapshot |
UTT5OTyIWm | FOUNDER: Grounding Foundation Models in World Models for Open-Ended Embodied Decision Making | https://openreview.net/forum?id=UTT5OTyIWm | [
"Yucen Wang",
"Rui Yu",
"Shenghua Wan",
"Le Gan",
"De-Chuan Zhan"
] | Poster | reinforcement_learning->deep_rl | Foundation Models (FMs) and World Models (WMs) offer complementary strengths in task generalization at different levels. In this work, we propose FOUNDER, a framework that integrates the generalizable knowledge embedded in FMs with the dynamic modeling capabilities of WMs to enable open-ended task solving in embodied e... | [
"Reinforcement Learning",
"World Models",
"Foundation Models",
"Open-ended Tasks",
"Goal-Conditioned Reinforcement Learning",
"Multi-task Reinforcement Learning",
"Learning Reward Functions"
] | We propose FOUNDER that grounds Foundation Model task representations into World Model goal states, enabling open-ended task specification and completion in embodied environments by capturing deep-level task semantics. | 15,878 | 2507.12496 | title_snapshot |
XfjrLEPOQV | Understanding Sharpness Dynamics in NN Training with a Minimalist Example: The Effects of Dataset Difficulty, Depth, Stochasticity, and More | https://openreview.net/forum?id=XfjrLEPOQV | [
"Geonhui Yoo",
"Minhak Song",
"Chulhee Yun"
] | Poster | deep_learning->theory | When training deep neural networks with gradient descent, sharpness often increases---a phenomenon known as *progressive sharpening*---before saturating at the *edge of stability*. Although commonly observed in practice, the underlying mechanisms behind progressive sharpening remain poorly understood. In this work, we ... | [
"progressive sharpening",
"sharpness"
] | We propose a minimalist model that successfully replicates the progressive sharpening and edge of stability phenomena, and empirically and theoretically analyze the effect of problem parameters in progressive sharpening | 15,877 | 2506.06940 | title_snapshot |
cUNfm13VUR | Comparing Few to Rank Many: Active Human Preference Learning Using Randomized Frank-Wolfe Method | https://openreview.net/forum?id=cUNfm13VUR | [
"Kiran Koshy Thekumparampil",
"Gaurush Hiranandani",
"Kousha Kalantari",
"Shoham Sabach",
"Branislav Kveton"
] | Poster | general_machine_learning->online_learning_active_learning_and_bandits | We study learning human preferences from limited comparison feedback, a core machine learning problem that is at the center of reinforcement learning from human feedback (RLHF). We formulate the problem as learning a Plackett-Luce (PL) model from a limited number of $K$-subset comparisons over a universe of $N$ items, ... | [
"active learning",
"human preference learning",
"comparison feedback",
"optimal design",
"Frank-Wolfe method"
] | We propose an efficient active learning algorithm for learning human preferences from K-way comparison of a large number of choices. | 15,875 | 2412.19396 | title_judge |
xYtLsWiUli | Vector Grimoire: Codebook-based Shape Generation under Raster Image Supervision | https://openreview.net/forum?id=xYtLsWiUli | [
"Marco Cipriano",
"Moritz Feuerpfeil",
"Gerard de Melo"
] | Poster | deep_learning->generative_models_and_autoencoders | Scalable Vector Graphics (SVG) is a popular format on the web and in the design industry. However, despite the great strides made in generative modeling, SVG has remained underexplored due to the discrete and complex nature of such data. We introduce GRIMOIRE, a text-guided SVG generative model that is comprised of two... | [
"Image Generation",
"Scalable Vector Graphics",
"VQ-VAE",
"Differentiable Rasterizer"
] | We propose a text-guided SVG generative model that creates vector graphics from natural language descriptions, learning from raster images without direct SVG supervision | 15,874 | 2410.05991 | title_snapshot |
WxY61MmHYo | Scaling Laws for Task-Optimized Models of the Primate Visual Ventral Stream | https://openreview.net/forum?id=WxY61MmHYo | [
"Abdulkadir Gokce",
"Martin Schrimpf"
] | Spotlight | applications->neuroscience_cognitive_science | When trained on large-scale object classification datasets, certain artificial neural network models begin to approximate core object recognition behaviors and neural response patterns in the primate brain. While recent machine learning advances suggest that scaling compute, model size, and dataset size improves task p... | [
"scaling laws",
"neural alignment",
"behavioral alignment",
"computer vision",
"primate visual ventral stream"
] | We systematically explored scaling laws for primate vision models and discovered that neural alignment stops improving beyond a certain scale, even though behavior keeps aligning better. | 15,849 | 2411.05712 | title_snapshot |
2dqiqST8ZJ | Score-Based Diffusion Policy Compatible with Reinforcement Learning via Optimal Transport | https://openreview.net/forum?id=2dqiqST8ZJ | [
"Mingyang Sun",
"Pengxiang Ding",
"Weinan Zhang",
"Donglin Wang"
] | Poster | reinforcement_learning->deep_rl | Diffusion policies have shown promise in learning complex behaviors from demonstrations, particularly for tasks requiring precise control and long-term planning. However, they face challenges in robustness when encountering distribution shifts. This paper explores improving diffusion-based imitation learning models thr... | [
"Diffusion Policy",
"Reinforcement Learning",
"Optimal Transport"
] | null | 15,847 | 2502.12631 | title_snapshot |
UNrfYfbLZ3 | Accelerating Linear Recurrent Neural Networks for the Edge with Unstructured Sparsity | https://openreview.net/forum?id=UNrfYfbLZ3 | [
"Alessandro Pierro",
"Steven Abreu",
"Jonathan Timcheck",
"Philipp Stratmann",
"Andreas Wild",
"Sumit Bam Shrestha"
] | Poster | general_machine_learning->hardware_and_software | Linear recurrent neural networks enable powerful long-range sequence modeling with constant memory usage and time-per-token during inference. These architectures hold promise for streaming applications at the edge, but deployment in resource-constrained environments requires hardware-aware optimizations to minimize lat... | [
"Linear RNNs",
"Sparsity",
"Pruning",
"Quantization",
"Neuromorphic Hardware"
] | null | 15,842 | 2502.01330 | title_snapshot |
iwkCnlOa2A | Complex Wavelet Mutual Information Loss: A Multi-Scale Loss Function for Semantic Segmentation | https://openreview.net/forum?id=iwkCnlOa2A | [
"Renhao Lu"
] | Poster | applications->computer_vision | Recent advancements in deep neural networks have significantly enhanced the performance of semantic segmentation. However, class imbalance and instance imbalance remain persistent challenges, where smaller instances and thin boundaries are often overshadowed by larger structures. To address the multiscale nature of seg... | [
"Semantic segmentation",
"wavelet transform",
"steerable pyramid",
"mutual information"
] | We propose a novel loss function, CWMI, that enhances image segmentation by combining multi-scale wavelet analysis with mutual information, achieving better edge precision and robustness without extra computation. | 15,833 | 2502.00563 | title_snapshot |
1d1ssNedLv | Balancing Model Efficiency and Performance: Adaptive Pruner for Long-tailed Data | https://openreview.net/forum?id=1d1ssNedLv | [
"Zhe Zhao",
"HaiBin Wen",
"Pengkun Wang",
"ShuangWang",
"Zhenkun Wang",
"Qingfu Zhang",
"Yang Wang"
] | Poster | general_machine_learning->representation_learning | Long-tailed distribution datasets are prevalent in many machine learning tasks, yet existing neural network models still face significant challenges when handling such data. This paper proposes a novel adaptive pruning strategy, LTAP (Long-Tailed Adaptive Pruner), aimed at balancing model efficiency and performance to ... | [
"Neural network pruning，Long-tail learning"
] | null | 15,821 | null | null |
ftR9OuiUJA | Contour Integration Underlies Human-Like Vision | https://openreview.net/forum?id=ftR9OuiUJA | [
"Ben Lonnqvist",
"Elsa Scialom",
"Abdulkadir Gokce",
"Zehra Merchant",
"Michael Herzog",
"Martin Schrimpf"
] | Poster | applications->neuroscience_cognitive_science | Despite the tremendous success of deep learning in computer vision, models still fall behind humans in generalizing to new input distributions. Existing benchmarks do not investigate the specific failure points of models by analyzing performance under many controlled conditions. Our study systematically dissects where ... | [
"psychophysics",
"machine vision",
"human vision",
"contour integration",
"robustness",
"visual perception"
] | Models fail at human-like contour integration; those that integrate contours in a more human-like way also reach better accuracy and robustness. | 15,810 | 2504.05253 | title_snapshot |
NOV32X1Rq3 | Towards Universal Offline Black-Box Optimization via Learning Language Model Embeddings | https://openreview.net/forum?id=NOV32X1Rq3 | [
"Rong-Xi Tan",
"Ming Chen",
"Ke Xue",
"Yao Wang",
"Yaoyuan Wang",
"Fu Sheng",
"Chao Qian"
] | Poster | optimization->zeroorder_and_blackbox_optimization | The pursuit of universal black-box optimization (BBO) algorithms is a longstanding goal. However, unlike domains such as language or vision, where scaling structured data has driven generalization, progress in offline BBO remains hindered by the lack of unified representations for heterogeneous numerical spaces. Thus, ... | [
"Universal optimization; Offline optimization"
] | null | 15,805 | 2506.07109 | title_snapshot |
xQTSvP57C3 | Nonlinear transformers can perform inference-time feature learning | https://openreview.net/forum?id=xQTSvP57C3 | [
"Naoki Nishikawa",
"Yujin Song",
"Kazusato Oko",
"Denny Wu",
"Taiji Suzuki"
] | Poster | theory->learning_theory | Pretrained transformers have demonstrated the ability to implement various algorithms at inference time without parameter updates. While theoretical works have established this capability through constructions and approximation guarantees, the optimization and statistical efficiency aspects remain understudied. In this... | [
"transformers",
"in-context learning",
"feature learning",
"single-index models"
] | null | 15,804 | null | null |
R65zHNqND0 | Bongard in Wonderland: Visual Puzzles that Still Make AI Go Mad? | https://openreview.net/forum?id=R65zHNqND0 | [
"Antonia Wüst",
"Tim Tobiasch",
"Lukas Helff",
"Inga Ibs",
"Wolfgang Stammer",
"Devendra Singh Dhami",
"Constantin A. Rothkopf",
"Kristian Kersting"
] | Poster | deep_learning->foundation_models | Recently, newly developed Vision-Language Models (VLMs), such as OpenAI's o1, have emerged, seemingly demonstrating advanced reasoning capabilities across text and image modalities. However, the depth of these advances in language-guided perception and abstract reasoning remains underexplored, and it is unclear whether... | [
"Visual Reasoning",
"Vision Language Models",
"Bongard problems"
] | Diagnostic evaluation of VLMs on Bongard Problems that reveals problems in perception and reasoning. | 15,802 | 2410.19546 | title_snapshot |
A82tIFgJaK | Harmonizing Geometry and Uncertainty: Diffusion with Hyperspheres | https://openreview.net/forum?id=A82tIFgJaK | [
"Muskan Dosi",
"Chiranjeev Chiranjeev",
"Kartik Thakral",
"Mayank Vatsa",
"Richa Singh"
] | Poster | deep_learning->generative_models_and_autoencoders | Do contemporary diffusion models preserve the class geometry of hyperspherical data? Standard diffusion models rely on isotropic Gaussian noise in the forward process, inherently favoring Euclidean spaces. However, many real-world problems involve non-Euclidean distributions, such as hyperspherical manifolds, where cla... | [
"Diffusion Model",
"vMF Distribution"
] | Exploring diffusion to generate class-aware hyperspherical geometry by leveraging angular uncertainty. | 15,800 | 2506.10576 | title_snapshot |
P0RkH1RT5z | Subgroups Matter for Robust Bias Mitigation | https://openreview.net/forum?id=P0RkH1RT5z | [
"Anissa Alloula",
"Charles Jones",
"Ben Glocker",
"Bartlomiej Papiez"
] | Poster | social_aspects->fairness | Despite the constant development of new bias mitigation methods for machine learning, no method consistently succeeds, and a fundamental question remains unanswered: when and why do bias mitigation techniques fail? In this paper, we hypothesise that a key factor may be the often-overlooked but crucial step shared by ma... | [
"bias mitigation",
"robustness",
"spurious correlations",
"generalisation",
"fairness"
] | Our work highlights the importance of careful subgroup definition in bias mitigation and suggest it as a alternative lever for improving the robustness and fairness of machine learning models. | 15,799 | 2505.21363 | title_snapshot |
UJXbcJ7qXB | Hyperbolic-PDE GNN: Spectral Graph Neural Networks in the Perspective of A System of Hyperbolic Partial Differential Equations | https://openreview.net/forum?id=UJXbcJ7qXB | [
"Juwei Yue",
"Haikuo Li",
"Jiawei Sheng",
"Xiaodong Li",
"Taoyu Su",
"Tingwen Liu",
"Li Guo"
] | Poster | deep_learning->graph_neural_networks | Graph neural networks (GNNs) leverage message passing mechanisms to learn the topological features of graph data. Traditional GNNs learns node features in a spatial domain unrelated to the topology, which can hardly ensure topological features. In this paper, we formulates message passing as a system of hyperbolic part... | [
"Graph Neural Networks",
"Hyperbolic Partial Differential Equations"
] | null | 15,796 | 2505.23014 | title_snapshot |
TV17MLZGuA | Mind the Gap: A Practical Attack on GGUF Quantization | https://openreview.net/forum?id=TV17MLZGuA | [
"Kazuki Egashira",
"Robin Staab",
"Mark Vero",
"Jingxuan He",
"Martin Vechev"
] | Poster | social_aspects->safety | With the increasing size of frontier LLMs, post-training quantization has become the standard for memory-efficient deployment. Recent work has shown that basic rounding-based quantization schemes pose security risks, as they can be exploited to inject malicious behaviors into quantized models that remain hidden in full... | [
"quantization",
"large language models",
"security",
"poisoning",
"gguf"
] | Building on existing LLM quantization exploitation attacks targeting naive quantization, we extend them to the popular GGUF quantization by a simple modification. | 15,795 | 2505.23786 | title_snapshot |
EIfCH9OgjR | Elucidating the design space of language models for image generation | https://openreview.net/forum?id=EIfCH9OgjR | [
"Xuantong LIU",
"Shaozhe Hao",
"Xianbiao Qi",
"Tianyang Hu",
"Jun Wang",
"Rong Xiao",
"Yuan Yao"
] | Poster | applications->computer_vision | The success of large language models (LLMs) in text generation has inspired their application to image generation. However, existing methods either rely on specialized designs with inductive biases or adopt LLMs without fully exploring their potential in vision tasks. In this work, we systematically investigate the des... | [
"Image generation",
"Large language model",
"Generative model"
] | This work fully explores the use of language models for image generation, analyzing their optimization behavior, investigating tokenization, sampling strategies, and model scalability to achieve optimal performance. | 15,791 | 2410.16257 | title_snapshot |
aTQtGq7IyT | Be a Goldfish: Forgetting Bad Conditioning in Sparse Linear Regression via Variational Autoencoders | https://openreview.net/forum?id=aTQtGq7IyT | [
"Kuheli Pratihar",
"Debdeep Mukhopadhyay"
] | Poster | deep_learning->generative_models_and_autoencoders | Variational Autoencoders (VAEs), a class of latent-variable generative models, have seen extensive use in high-fidelity synthesis tasks, yet their loss landscape remains poorly understood. Prior theoretical works on VAE loss analysis have focused on their latent-space representational capabilities, both in the optimal ... | [
"Variational Autoencoders",
"Sparse Linear Regression",
"NP-Hard",
"Smoothing",
"Sparsity",
"Matrix Conditioning"
] | We use Variational Autoencoders that smoothen out bad local minima to solve the NP-hard inverse problem of Sparse linear regression and perform better than conventional methods. | 15,780 | null | null |
LiXD7mpjU0 | Incremental Gradient Descent with Small Epoch Counts is Surprisingly Slow on Ill-Conditioned Problems | https://openreview.net/forum?id=LiXD7mpjU0 | [
"Yujun Kim",
"Jaeyoung Cha",
"Chulhee Yun"
] | Poster | theory->optimization | Recent theoretical results demonstrate that the convergence rates of permutation-based SGD (e.g., random reshuffling SGD) are faster than uniform-sampling SGD; however, these studies focus mainly on the large epoch regime, where the number of epochs $K$ exceeds the condition number $\kappa$. In contrast, little is know... | [
"Permutation-based SGD",
"Incremental Gradient Descent",
"Lower Bound",
"Convex Optimization"
] | We study Incremental Gradient Descent in the small epoch regime and show that it exhibits severe slowdown especially in the presence of nonconvex components. | 15,777 | 2506.04126 | title_snapshot |
9rLxi2cnZC | Lightweight Dataset Pruning without Full Training via Example Difficulty and Prediction Uncertainty | https://openreview.net/forum?id=9rLxi2cnZC | [
"Yeseul Cho",
"Baekrok Shin",
"Changmin Kang",
"Chulhee Yun"
] | Poster | general_machine_learning->supervised_learning | Recent advances in deep learning rely heavily on massive datasets, leading to substantial storage and training costs. Dataset pruning aims to alleviate this demand by discarding redundant examples. However, many existing methods require training a model with a full dataset over a large number of epochs before being abl... | [
"Dataset Pruning",
"Coreset Selection",
"Example Difficulty",
"Prediction Uncertainty"
] | For dataset pruning, we introduce the DUAL (Difficulty and Uncertainty-Aware Lightweight) score, a new method that identifies important training examples early in the training process by considering both example difficulty and prediction uncertainty. | 15,764 | 2502.06905 | title_snapshot |
SY4owu5BK6 | The Case for Learned Provenance-based System Behavior Baseline | https://openreview.net/forum?id=SY4owu5BK6 | [
"Yao Zhu",
"Zhenyuan LI",
"Yangyang Wei",
"Shouling Ji"
] | Poster | applications->everything_else | Provenance graphs describe data flows and causal dependencies of host activities, enabling to track the data propagation and manipulation throughout the systems, which provide a foundation for intrusion detection. However, these Provenance-based Intrusion Detection Systems (PIDSes) face significant challenges in storag... | [
"graph representation learning",
"provenance graph",
"cyber attack detection"
] | null | 15,752 | null | null |
ba3sSfEnj1 | Flexible, Efficient, and Stable Adversarial Attacks on Machine Unlearning | https://openreview.net/forum?id=ba3sSfEnj1 | [
"Zihan Zhou",
"Yang Zhou",
"Zijie Zhang",
"Lingjuan Lyu",
"Da Yan",
"Ruoming Jin",
"Dejing Dou"
] | Poster | general_machine_learning->everything_else | Machine unlearning (MU) aims to remove the influence of specific data points from trained models, enhancing compliance with privacy regulations. However, the vulnerability of basic MU models to malicious unlearning requests in adversarial learning environments has been largely overlooked. Existing adversarial MU attack... | [
"Machine unlearning",
"poisoning attack",
"thrust vector control theory",
"John's Theorem",
"polyhedral approximation"
] | Flexible, Efficient, and Stable Adversarial Attacks against machine unlearning that processes multiple arbitrary attack targets at a time | 15,747 | null | null |
O0lxLP4ABD | PipeOffload: Improving Scalability of Pipeline Parallelism with Memory Optimization | https://openreview.net/forum?id=O0lxLP4ABD | [
"Xinyi Wan",
"Penghui Qi",
"Guangxing Huang",
"Min Lin",
"Jialin Li"
] | Poster | optimization->large_scale_parallel_and_distributed | Pipeline parallelism (PP) is widely used for training large language models (LLMs), yet its scalability is often constrained by high activation memory consumption as the number of in-flight microbatches grows with the degree of PP. In this paper, we focus on addressing this challenge by leveraging the under-explored me... | [
"LLM Training",
"Pipeline Parallelism",
"Offloading",
"Memory Optimization"
] | null | 15,731 | 2503.01328 | title_snapshot |
z4XS0Ie391 | Unified Screening for Multiple Diseases | https://openreview.net/forum?id=z4XS0Ie391 | [
"Yiğit Narter",
"Alihan Hüyük",
"Mihaela van der Schaar",
"Cem Tekin"
] | Poster | applications->health_medicine | Current screening programs that focus on improving patient health while minimizing screening costs are tailored for individual diseases. Designing unified screening programs for multiple diseases requires carefully balancing competing disease risks, which is an open problem. In this work, we address this problem by cas... | [
"Unified screening",
"healthcare",
"competing risks",
"constrained optimization",
"threshold policy"
] | null | 15,730 | null | null |
39JKH8k3FS | Causal Effect Identification in lvLiNGAM from Higher-Order Cumulants | https://openreview.net/forum?id=39JKH8k3FS | [
"Daniele Tramontano",
"Yaroslav Kivva",
"Saber Salehkaleybar",
"Negar Kiyavash",
"Mathias Drton"
] | Poster | general_machine_learning->causality | This paper investigates causal effect identification in latent variable Linear Non-Gaussian Acyclic Models (lvLiNGAM) using higher-order cumulants, addressing two prominent setups that are challenging in the presence of latent confounding: (1) a single proxy variable that may causally influence the treatment and (2) un... | [
"Causal Effect Identification",
"Structural Causal Models",
"Cumulants."
] | null | 15,727 | 2506.05202 | title_snapshot |
qF6mxani2X | STAMP Your Content: Proving Dataset Membership via Watermarked Rephrasings | https://openreview.net/forum?id=qF6mxani2X | [
"Saksham Rastogi",
"Pratyush Maini",
"Danish Pruthi"
] | Poster | social_aspects | Given how large parts of publicly available text are crawled to pretrain large language models (LLMs), data creators increasingly worry about the inclusion of their proprietary data for model training without attribution or licensing. Their concerns are also shared by benchmark curators whose test-sets might be comprom... | [
"LLM",
"membership inference",
"dataset inference",
"watermarking",
"test set contamination"
] | We present STAMP, a framework to detect whether a given dataset was used in LLM pretraining. | 15,721 | 2504.13416 | title_snapshot |
jHLSnYNt1m | Counterfactual Effect Decomposition in Multi-Agent Sequential Decision Making | https://openreview.net/forum?id=jHLSnYNt1m | [
"Stelios Triantafyllou",
"Aleksa Sukovic",
"Yasaman Zolfimoselo",
"Goran Radanovic"
] | Poster | general_machine_learning->causality | We address the challenge of explaining counterfactual outcomes in multi-agent Markov decision processes. In particular, we aim to explain the total counterfactual effect of an agent's action on the outcome of a realized scenario through its influence on the environment dynamics and the agents' behavior. To achieve this... | [
"counterfactual reasoning",
"causal explanation formula",
"multi-agent Markov decision processes",
"accountability"
] | We address the challenge of explaining the total counterfactual effect of an agent’s action on the outcome of a realized scenario in multi-agent Markov decision processes. | 15,708 | 2410.12539 | title_snapshot |
JsPyLqCgks | A Mixed-Curvature based Pre-training Paradigm for Multi-Task Vehicle Routing Solver | https://openreview.net/forum?id=JsPyLqCgks | [
"Suyu Liu",
"Zhiguang Cao",
"Shanshan Feng",
"Yew-Soon Ong"
] | Poster | optimization->discrete_and_combinatorial_optimization | Solving various types of vehicle routing problems (VRPs) using a unified neural solver has garnered significant attentions in recent years. Despite their effectiveness, existing neural multi-task solvers often fail to account for the geometric structures inherent in different tasks, which may result in suboptimal perfo... | [
"Combinatorial Optimization",
"Constrained Optimization",
"Vehicle Routing Problems",
"Deep Reinforcement Learning"
] | We propose a geometric pre-training method to get a more powerful foundation model for various kinds of VRPs. | 15,699 | null | null |
rxKC8v2uHc | GRAM: A Generative Foundation Reward Model for Reward Generalization | https://openreview.net/forum?id=rxKC8v2uHc | [
"Chenglong Wang",
"Yang Gan",
"Yifu Huo",
"Yongyu Mu",
"Qiaozhi He",
"MuRun Yang",
"Bei Li",
"Tong Xiao",
"Chunliang Zhang",
"Tongran Liu",
"JingBo Zhu"
] | Poster | deep_learning->large_language_models | In aligning large language models (LLMs), reward models have played an important role, but are standardly trained as discriminative models and rely only on labeled human preference data. In this paper, we explore methods that train reward models using both unlabeled and labeled data. Building on the generative models... | [
"Large Language Model",
"Reward Modeling",
"RLHF"
] | This study introduces a generative foundation reward model (GRAM), which pre-learns preferences through a two-stage training process and selective label smoothing. | 15,682 | 2506.14175 | title_snapshot |
aTC2euLwnh | Fine-Grained Captioning of Long Videos through Scene Graph Consolidation | https://openreview.net/forum?id=aTC2euLwnh | [
"Sanghyeok Chu",
"Seonguk Seo",
"Bohyung Han"
] | Poster | applications->computer_vision | Recent advances in vision-language models have led to impressive progress in caption generation for images and short video clips. However, these models remain constrained by their limited temporal receptive fields, making it difficult to produce
coherent and comprehensive captions for long videos. While several methods... | [
"Long video captioning",
"zero-shot video captioning",
"scene graph"
] | We propose a novel framework for long video captioning based on graph consolidation. | 15,674 | 2502.16427 | title_snapshot |
tYwKQMMjJA | M3-JEPA: Multimodal Alignment via Multi-gate MoE based on the Joint-Embedding Predictive Architecture | https://openreview.net/forum?id=tYwKQMMjJA | [
"Hongyang Lei",
"Xiaolong Cheng",
"Qi Qin",
"Dan Wang",
"Huazhen Huang",
"Qingqing Gu",
"Yetao Wu",
"Luo Ji"
] | Poster | deep_learning->selfsupervised_learning | Current multimodal learning strategies primarily optimize in the original token space. Such a framework is easy to incorporate with the backbone of pretrained language model, but might result in modality collapse. To alleviate such issues, we leverage the Joint-Embedding Predictive Architecture (JEPA) on the multimodal... | [
"JEPA",
"MoE",
"multimodal",
"alignment"
] | We propose M3-JEPA, which implements a multi-gate MoE predictor to JEPA, and shows theoretical optimality and reasonable performance on multimodal tasks.. | 15,673 | 2409.05929 | title_snapshot |
qzM37nOy3N | Inverse problems with experiment-guided AlphaFold | https://openreview.net/forum?id=qzM37nOy3N | [
"Sai Advaith Maddipatla",
"Nadav Bojan",
"Meital Bojan",
"Sanketh Vedula",
"Paul Schanda",
"Ailie Marx",
"Alexander Bronstein"
] | Poster | applications->chemistry_physics_and_earth_sciences | Proteins exist as a dynamic ensemble of multiple conformations, and these motions are often crucial for their functions. However, current structure prediction methods predominantly yield a single conformation, overlooking the conformational heterogeneity revealed by diverse experimental modalities. Here, we present a ... | [
"protein structure prediction",
"alphafold",
"protein generative models",
"experiment-grounded generative models"
] | We develop experiment-guided AlphaFold-3 to solve inverse problems in structural biology, leading to faster experimental cycles and improved modeling in X-ray crystallography and NMR spectroscopy. | 15,661 | 2502.09372 | title_snapshot |
P9DQ2IExgS | Synthesizing Software Engineering Data in a Test-Driven Manner | https://openreview.net/forum?id=P9DQ2IExgS | [
"Lei Zhang",
"Jiaxi Yang",
"Min Yang",
"Jian Yang",
"Mouxiang Chen",
"Jiajun Zhang",
"Zeyu Cui",
"Binyuan Hui",
"Junyang Lin"
] | Poster | applications | We introduce **SWE-Flow**, a novel data synthesis framework grounded in Test-Driven Development (TDD).
Unlike existing software engineering data that rely on human-submitted issues, **SWE-Flow** automatically infers incremental development steps directly from unit tests, which inherently encapsulate high-level requirem... | [
"Large language Model",
"Software Engineering",
"Test-Driven Development",
"Code Agent"
] | null | 15,658 | 2506.09003 | title_judge |
mQE0EsrX1y | AEQA-NAT : Adaptive End-to-end Quantization Alignment Training Framework for Non-autoregressive Machine Translation | https://openreview.net/forum?id=mQE0EsrX1y | [
"Xiangyu Qu",
"guojing liu",
"Liang Li"
] | Poster | deep_learning->sequential_models_time_series | Non-autoregressive Transformers (NATs) have garnered significant attention due to their efficient decoding compared to autoregressive methods. However, existing conditional dependency modeling schemes based on masked language modeling introduce a *training-inference gap* in NATs. For instance, while NATs sample target ... | [
"Machine Translation",
"Parallel decoding",
"Vector Quantization"
] | null | 15,652 | null | null |
cYNBsMTAVL | Portable Reward Tuning: Towards Reusable Fine-Tuning across Different Pretrained Models | https://openreview.net/forum?id=cYNBsMTAVL | [
"Daiki Chijiwa",
"Taku Hasegawa",
"Kyosuke Nishida",
"Kuniko Saito",
"Susumu Takeuchi"
] | Poster | deep_learning->foundation_models | While foundation models have been exploited for various expert tasks with their fine-tuned parameters, any foundation model will be eventually outdated due to its old knowledge or limited capability, and thus should be replaced by a new foundation model. Subsequently, to benefit from its latest knowledge or improved ca... | [
"inference-time tuning",
"reward maximization"
] | null | 15,647 | 2502.12776 | title_snapshot |
CXPpYJpYXQ | LOB-Bench: Benchmarking Generative AI for Finance - an Application to Limit Order Book Data | https://openreview.net/forum?id=CXPpYJpYXQ | [
"Peer Nagy",
"Sascha Yves Frey",
"Kang Li",
"Bidipta Sarkar",
"Svitlana Vyetrenko",
"Stefan Zohren",
"Ani Calinescu",
"Jakob Nicolaus Foerster"
] | Poster | applications->time_series | While financial data presents one of the most challenging and interesting sequence modelling tasks due to high noise, heavy tails, and strategic interactions, progress in this area has been hindered by the lack of consensus on quantitative evaluation paradigms.
To address this, we present **LOB-Bench**, a benchmark, i... | [
"finance",
"generative models",
"time series",
"state-space models",
"benchmark"
] | LOB-Bench offers a rigorous framework and open-source Python package for standardized evaluation of generative limit order book data models, addressing evaluation gaps and enhancing model comparisons with quantitative metrics. | 15,644 | 2502.09172 | title_snapshot |
End of preview. Expand in Data Studio
README.md exists but content is empty.
- Downloads last month
- 91