2278 papers
This paper introduces AgarCL, a research platform based on the non-episodic game Agar.io designed to advance continual reinforcement learning by providing a complex, dynamic environment where standard algorithms and existing continual learning methods face significant challenges beyond the traditional stability-plasticity dilemma.
This paper establishes a maximum principle and the Hamilton-Jacobi-Bellman equation for optimal control on infinite-dimensional probability distribution spaces, and leverages these theoretical results to develop a scalable deep learning algorithm for solving high-dimensional multi-agent control problems.
This paper proposes a novel defense against prompt injection attacks in large language models by augmenting intermediate token representations with layer-specific trainable embeddings to enforce instruction hierarchy, achieving a 1.6x to 9.2x reduction in attack success rates compared to state-of-the-art methods without compromising model utility.
This paper proposes Orthogonal Common Neighbor (OCN), a novel link prediction method that addresses redundancy and over-smoothing in higher-order common neighbors through orthogonalization and normalization, achieving significant performance improvements over state-of-the-art baselines.
This paper introduces a novel representation of local protein environments derived from atomistic foundation models that effectively captures structural and chemical features, enabling the construction of data-driven priors and achieving state-of-the-art accuracy in physics-informed NMR chemical shift prediction.
This paper introduces MMTU, a large-scale benchmark comprising over 28,000 questions across 25 real-world expert-level table tasks, designed to comprehensively evaluate and reveal the significant limitations of current frontier models in understanding, reasoning, and manipulating structured tabular data.
BemaGANv2 is an advanced GAN-based vocoder that enhances long-term audio generation for Text-to-Music and Text-to-Audio applications by integrating Anti-aliased Multi-Periodicity composition modules in the generator and systematically evaluating novel discriminator combination strategies, including the Multi-Envelope Discriminator, to achieve high-fidelity and temporally coherent results.
This paper introduces Co-LoRA, a collaborative personalization framework that addresses both data and model heterogeneity through a task-relevance-aware aggregation strategy and a dimension-invariant module, validated by a new multi-modal benchmark and superior performance over state-of-the-art methods.
This paper proposes a semi-self-supervised learning approach featuring a novel GLMask representation and a semantic-to-instance pipeline that achieves state-of-the-art instance segmentation performance with minimal manual annotation, demonstrating superior results on both dense agricultural wheat head images and the general-purpose COCO dataset.
This paper introduces SamS, an efficient algorithm that adaptively schedules training samples in Direct Preference Optimization based on the model's evolving batch-wise states, significantly improving LLM alignment performance without modifying the core DPO algorithm or incurring substantial computational overhead.
This paper introduces MeRF, a method that enhances reinforcement finetuning of large reasoning models by injecting reward specifications directly into prompts as "motivation," thereby leveraging in-context learning to align generation with optimization objectives and achieve substantial performance gains over standard RLVR baselines.
The paper proposes SUBARU, a power-efficient framework for hearables that intentionally employs sub-Nyquist sampling and low bit-resolution ADCs to achieve a 3.31x reduction in power consumption while maintaining high-quality multimodal speech enhancement through a novel wideband reconstruction methodology.
This paper establishes that Physics-Informed Neural Networks (PINNs) require a network size scaling with the number of noisy samples to achieve empirical risk below the noise variance, demonstrating that simply increasing data quantity cannot compensate for insufficient model capacity in noisy PDE training.
This paper identifies a pervasive "agreement bias" in Multimodal LLM verifiers that causes them to over-validate agent behavior, and proposes a lightweight Self-Grounded Verification (SGV) method that significantly improves failure detection and task completion across web navigation, computer use, and robotics by decoupling prior generation from trajectory evaluation.
This paper presents the first comprehensive survey of flow matching applications in biology and life sciences, systematically reviewing its theoretical foundations and categorizing its recent advancements in biological sequence modeling, molecule design, and protein generation.
This paper introduces User Goal State Tracking (UGST), a novel framework and three-stage methodology that enables LLM-based user simulators to autonomously track goal progression and generate goal-aligned responses, significantly improving performance on MultiWOZ 2.4 and -Bench benchmarks.
The paper introduces CauKer, a novel algorithm that combines Gaussian Process kernel composition with Structural Causal Models to generate diverse, causally coherent synthetic time series, enabling sample-efficient pre-training of classification foundation models that exhibit clear scaling laws across varying dataset sizes and model capacities.
GraphProp is a two-phase framework for training graph foundation models that first learns structural generalization by predicting graph invariants and then leverages these representations as positional encodings to enhance cross-domain performance in graph-level tasks, particularly outperforming existing methods in scenarios with limited data or missing node attributes.
The paper introduces ECHO, a novel foundation model that leverages band-split architecture and frequency positional embeddings to achieve state-of-the-art performance in anomaly detection and fault classification across variable-length, arbitrary sampling rate machine signals without requiring padding or cropping.
This paper proposes a unified training framework that combines entropy-driven curriculum learning, which sequences training from simple to complex trajectories based on Lempel-Ziv compression, with multi-task learning to simultaneously optimize location, distance, and direction predictions, thereby achieving state-of-the-art performance and significantly faster convergence in human mobility prediction.