🎯 What it does: A data-forgetfulness-based alignment method VAA is proposed, which balances forgettable and memorable samples through group learning to enhance robustness against harmful fine-tuning.
Watch Out Your Album! On the Inadvertent Privacy Memorization in Multi-Modal Large Language Models
Tianjie Ju (Shanghai Jiao Tong University), Gongshen Liu (Shanghai Jiao Tong University)
CodeSafty and PrivacyTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelMultimodality
🎯 What it does: This study investigates whether multimodal large models unintentionally memorize privacy content unrelated to tasks during the fine-tuning process and proposes a detection method based on gradient similarity and hierarchical probing.
🎯 What it does: A weakly supervised contrastive learning framework is proposed, which constructs continuous positive and negative samples through semantic similarity, effectively utilizing imprecise labels.
Weight matrices compression based on PDB model in deep neural networks
Xiaoling Wu (Southern University of Science and Technology), Zeng Li (Southern University of Science and Technology)
CodeCompressionImageText
🎯 What it does: This paper proposes the Population Double Bulk (PDB) model and the corresponding weight matrix compression algorithm, which can automatically determine the boundary between noise and information and compress the network without introducing hyperparameters.
🎯 What it does: This paper studies the Lottery Ticket Hypothesis (LTH) in sparse initialized Graph Neural Networks (GNNs), proving that it is still possible to find trainable sparse sub-networks while maintaining sufficient expressive power (i.e., comparable to 1-WL);
What Makes In-context Learning Effective for Mathematical Reasoning
Jiayu Liu (University of Science and Technology of China), Enhong Chen (University of Science and Technology of China)
CodeTransformerLarge Language ModelText
🎯 What it does: This paper quantifies the impact of examples on the mathematical reasoning performance of large language models in context learning through theoretical analysis and proposes an example selection method called LMS3 based on semantic similarity and reasoning stability.
When can in-context learning generalize out of task distribution?
Chase Goddard (Princeton University), David J. Schwab (Graduate Center CUNY)
CodeTransformerLarge Language ModelTabular
🎯 What it does: This paper studies whether the Transformer can generalize to out-of-distribution tasks when the diversity of pre-training task distributions changes. By sampling linear regression tasks on high-dimensional spherical sections at different angles ϕ and training a GPT-2 style Transformer for in-context learning (ICL), a transition from 'specialized' solutions to 'general' solutions was observed; this transition was further validated in classification and nonlinear regression tasks.
🎯 What it does: This paper studies the reasons why Maximum Entropy Reinforcement Learning (MaxEnt RL) may lead to misleading policy optimization in complex control tasks, and proposes the 'Entropy Bifurcation Extension' method to demonstrate the existence of this misleading effect. It further experimentally verifies the reasons for the poor performance of Soft Actor-Critic (SAC) in real control environments (such as high-speed vehicles, quadcopters, and quadruped robots) and explores the mitigating effects of adaptive entropy regulation.
🎯 What it does: The study investigates the impact of confounding factors on model performance in continuous learning scenarios, proposing the ConCon dataset and evaluating various CL methods.
🎯 What it does: Proposes the WikiBigEdit benchmark, providing over 500,000 real-time knowledge editing question-answer data based on Wikidata, supporting large-scale lifelong knowledge editing research.
Windows Agent Arena: Evaluating Multi-Modal OS Agents at Scale
Rogerio Bonatti (Microsoft), Zheng Hui
CodeTransformerLarge Language ModelAgentic AITextMultimodalityBenchmarkChain-of-Thought
🎯 What it does: Introducing Windows Agent Arena, a scalable multimodal agent benchmark built on a real Windows 11 environment, containing 154 tasks across 11 categories including office, browsing, system, programming, and media.
Winner-takes-all for Multivariate Probabilistic Time Series Forecasting
Adrien Cortes (Institut Polytechnique de Paris), Victor Letzelter (Valeo)
CodeRecurrent Neural NetworkTime SeriesFinance Related
🎯 What it does: The TimeMCL method is proposed, which utilizes multi-head neural networks and Winner-Takes-All (WTA) loss to generate multiple feasible future prediction paths with a single forward pass.
🎯 What it does: This paper proposes the Wolfpack adversarial attack strategy and its corresponding WALL robust training framework to enhance the reliability of multi-agent reinforcement learning in the face of coordinated attacks.
WOMD-Reasoning: A Large-Scale Dataset for Interaction Reasoning in Driving
Yiheng Li (University of California Berkeley), Wei Zhan
CodeAutonomous DrivingTransformerLarge Language ModelSupervised Fine-TuningVision Language ModelVideoTextMultimodalityChain-of-Thought
🎯 What it does: A large-scale interactive reasoning dataset, WOMD-Reasoning, has been constructed, and Motion-LLaVA has been fine-tuned based on it to enhance the understanding and prediction of traffic rules and human intention-driven interactions.
🎯 What it does: This paper proposes a discrete diffusion model based on Wyckoff positions, called WyckoffDiff, which can generate prototype structures that satisfy crystal symmetry constraints.
Rahul Sharma (Deutsches Forschungszentrum für Künstliche Intelligenz GmbH), David Antony Selby (Deutsches Forschungszentrum für Künstliche Intelligenz GmbH)
CodeOptimizationExplainability and InterpretabilityAuto EncoderTabular
🎯 What it does: This study proposes the concept of 'X-hacking' and explores how to use AutoML to find models that maintain predictive performance while providing pre-defined explanations (such as SHAP values) within model diversity (Rashomon phenomenon), which can intentionally or unintentionally mislead model interpretations.
🎯 What it does: Using multi-view 3D reconstruction to obtain voxel representation, robust texture optimization is applied to the color channel, significantly improving model recognition accuracy under various image degradation conditions.