We introduce Entropy-Guided Sequence Weighting (EGSW), a novel approach that enhances the exploration-exploitation tradeoff by dynamically assigning weights to generated outputs based on their advantage and entropy for Reinforcement Learning-based Large Language Model fine-tuning. EGSW integrates entropy regularization with advantage-based weighting to balance policy updates, enabling efficient exploration in high-dimensional state spaces. By employing temperature-scaled softmax weighting over sequences, EGSW prioritizing high-reward, high-uncertainty steps while maintaining training stability. Although originally developed to improve Group Relative Policy Optimization (GRPO) during large language model (LLM) fine-tuning, EGSW is generalizable to other reinforcement learning (RL) algorithms and can be implemented in both step-wise and trajectory-wise settings. Empirical evaluations demonstrate that EGSW enhances GRPO reasoning ability, yielding improvements in sample efficiency. Future work will explore the application of EGSW to advanced RL methodologies.

-- We introduce the sequential multi-object robotic grasp sampling algorithm SeqGrasp that can robustly synthesize stable grasps on diverse objects using the robotic hand's partial Degrees of Freedom (DoF). We use SeqGrasp to construct the large-scale Allegro Hand sequential grasping dataset SeqDataset and use it for training the diffusion-based sequential grasp generator SeqDiffuser . We experimentally evaluate SeqGrasp and SeqDiffuser against the state-of-the-art non-sequential multi-object grasp generation method Multi-Grasp in simulation and on a real robot. Furthermore, SeqDiffuser is approximately 1000 times faster at generating grasps than SeqGrasp and MultiGrasp. Generation of dexterous grasps has been studied for a long time, both from a technical perspective on generating grasps on robots [1]-[11] and understanding human grasping [12]- [15]. Most of these methods rely on bringing the robotic hand close to the object and then simultaneously enveloping it with all fingers. While this strategy often results in efficient and successful grasp generation, it simplifies dexterous grasping to resemble parallel-jaw grasping, thereby underutilizing the many DoF of multi-fingered robotic hands [10]. In contrast, grasping multiple objects with a robotic hand, particularly in a sequential manner that mirrors human-like dexterity, as shown in Figure 1, is still an unsolved problem. In this work, we introduce SeqGrasp, a novel hand-agnostic algorithm for generating sequential multi-object grasps.

The discovery of novel small molecule drugs remains a critical scientific challenge with far-reaching implications for treating diseases and advancing human health. Traditional drug development--especially for small molecule therapeutics--is a highly complex, resource-intensive, and time-consuming process that requires multidisciplinary collaboration. Recent breakthroughs in artificial intelligence (AI), particularly the rise of large language models (LLMs), present a transformative opportunity to streamline and accelerate this process. In this paper, we introduce PharmAgents, a virtual pharmaceutical ecosystem driven by LLM-based multi-agent collaboration. PharmAgents simulates the full drug discovery workflow--from target discovery to preclinical evaluation--by integrating explainable, LLM-driven agents equipped with specialized machine learning models and computational tools. Through structured knowledge exchange and automated optimization, PharmAgents identifies potential therapeutic targets, discovers promising lead compounds, enhances binding affinity and key molecular properties, and performs in silico analyses of toxicity and synthetic feasibility. Additionally, the system supports interpretability, agent interaction, and self-evolvement, enabling it to refine future drug designs based on prior experience. By showcasing the potential of LLM-powered multi-agent systems in drug discovery, this work establishes a new paradigm for autonomous, explainable, and scalable pharmaceutical research, with future extensions toward comprehensive drug lifecycle management.

In continual learning (CL), catastrophic forgetting often arises due to feature drift. This challenge is particularly prominent in the exemplar-free continual learning (EFCL) setting, where samples from previous tasks cannot be retained, making it difficult to preserve prior knowledge. To address this issue, some EFCL methods aim to identify feature spaces that minimize the impact on previous tasks while accommodating new ones. However, they rely on static features or outdated statistics stored from old tasks, which prevents them from capturing the dynamic evolution of the feature space in CL, leading to performance degradation over time. In this paper, we introduce the Drift-Resistant Space (DRS), which effectively handles feature drifts without requiring explicit feature modeling or the storage of previous tasks. A novel parameter-efficient fine-tuning approach called Low-Rank Adaptation Subtraction (LoRA-) is proposed to develop the DRS. This method subtracts the LoRA weights of old tasks from the initial pre-trained weight before processing new task data to establish the DRS for model training. Therefore, LoRA- enhances stability, improves efficiency, and simplifies implementation. Furthermore, stabilizing feature drifts allows for better plasticity by learning with a triplet loss. Our method consistently achieves state-of-the-art results, especially for long task sequences, across multiple datasets.

--This paper explores the design of beamforming codebooks for the base station (BS) and for the reconfigurable intelligent surfaces (RISs) in an active sensing scheme for uplink localization, in which the mobile user transmits a sequence of pilots to the BS through reflection at the RISs, and the BS and the RISs are adaptively configured by carefully choosing BS beamforming codeword and RIS codewords from their respective codebooks in a sequential manner to progressively focus onto the user . Most existing codebook designs for RIS are not tailored for active sensing, by which we mean the choice of the next codeword should depend on the measurements made so far, and the sequence of codewords should dynamically focus reflection toward the user . Moreover, most existing codeword selection methods rely on exhaustive search in beam training to identify the codeword with the highest signal-to-noise ratio (SNR), thus incurring substantial pilot overhead as the size of the codebook scales. This paper proposes a learning-based approach for codebook construction and for codeword selection for active sensing. The proposed learning approach aims to locate a target in the service area by recursively selecting a sequence of BS beamforming codewords and RIS codewords from the respective codebooks as more measurements become available without exhaustive beam training. The codebook design and the codeword selection fuse key ideas from the vector quantized variational autoencoder (VQ-V AE) and the long short-term memory (LSTM) network to learn respectively the discrete function space of the codebook and the temporal dependencies between measurements. The device is typically placed in the reflecting path between the transceivers, with its configuration wirelessly controlled by the transceivers via a control link. Manuscript submitted to IEEE Transactions on Wireless Communications on September 6, 2024, revised on January 12, 2025, accepted on March 5, 2025. Wei Y u is with The Edward S. Rogers Sr. This work is supported by the Natural Sciences and Engineering Research Council of Canada via the Canada Research Chairs program. The materials in this paper have been accepted in part at the IEEE Workshop on Signal Processing Advances in Wireless Communications (SP A WC), Lucca, Italy, September 2024 [1]. Codebook-based limited control link rate protocol can substantially reduce the control overhead [7], [8]. With the RIS codebook stored at the controller and at the RIS, the controller only needs to send the codeword index in order to configure the RIS.

Designing efficient learning algorithms with complexity guarantees for Markov decision processes (MDPs) with large or continuous state and action spaces remains a fundamental challenge. We address this challenge for entropy-regularized MDPs with Polish state and action spaces, assuming access to a generative model of the environment. We propose a novel family of multilevel Monte Carlo (MLMC) algorithms that integrate fixed-point iteration with MLMC techniques and a generic stochastic approximation of the Bellman operator. We quantify the precise impact of the chosen approximate Bellman operator on the accuracy of the resulting MLMC estimator. Leveraging this error analysis, we show that using a biased plain MC estimate for the Bellman operator results in quasi-polynomial sample complexity, whereas an unbiased randomized multilevel approximation of the Bellman operator achieves polynomial sample complexity in expectation. Notably, these complexity bounds are independent of the dimensions or cardinalities of the state and action spaces, distinguishing our approach from existing algorithms whose complexities scale with the sizes of these spaces. We validate these theoretical performance guarantees through numerical experiments.

View-invariant representation learning from egocentric (first-person, ego) and exocentric (third-person, exo) videos is a promising approach toward generalizing video understanding systems across multiple viewpoints. However, this area has been underexplored due to the substantial differences in perspective, motion patterns, and context between ego and exo views. In this paper, we propose a novel masked ego-exo modeling that promotes both causal temporal dynamics and cross-view alignment, called Bootstrap Your Own Views (BYOV), for fine-grained view-invariant video representation learning from unpaired ego-exo videos. We highlight the importance of capturing the compositional nature of human actions as a basis for robust cross-view understanding. Specifically, self-view masking and cross-view masking predictions are designed to learn view-invariant and powerful representations concurrently. Experimental results demonstrate that our BYOV significantly surpasses existing approaches with notable gains across all metrics in four downstream ego-exo video tasks. The code is available at https://github.com/park-jungin/byov.

Malicious image manipulation poses societal risks, increasing the importance of effective image manipulation detection methods. Recent approaches in image manipulation detection have largely been driven by fully supervised approaches, which require labor-intensive pixel-level annotations. Thus, it is essential to explore weakly supervised image manipulation localization methods that only require image-level binary labels for training. However, existing weakly supervised image manipulation methods overlook the importance of edge information for accurate localization, leading to suboptimal localization performance. To address this, we propose a Context-Aware Boundary Localization (CABL) module to aggregate boundary features and learn context-inconsistency for localizing manipulated areas. Furthermore, by leveraging Class Activation Mapping (CAM) and Segment Anything Model (SAM), we introduce the CAM-Guided SAM Refinement (CGSR) module to generate more accurate manipulation localization maps. By integrating two modules, we present a novel weakly supervised framework based on a dual-branch Transformer-CNN architecture. Our method achieves outstanding localization performance across multiple datasets.

Deep neural networks learn structured features from complex, non-Gaussian inputs, but the mechanisms behind this process remain poorly understood. Our work is motivated by the observation that the first-layer filters learnt by deep convolutional neural networks from natural images resemble those learnt by independent component analysis (ICA), a simple unsupervised method that seeks the most non-Gaussian projections of its inputs. This similarity suggests that ICA provides a simple, yet principled model for studying feature learning. Here, we leverage this connection to investigate the interplay between data structure and optimisation in feature learning for the most popular ICA algorithm, FastICA, and stochastic gradient descent (SGD), which is used to train deep networks. We rigorously establish that FastICA requires at least $n\gtrsim d^4$ samples to recover a single non-Gaussian direction from $d$-dimensional inputs on a simple synthetic data model. We show that vanilla online SGD outperforms FastICA, and prove that the optimal sample complexity $n \gtrsim d^2$ can be reached by smoothing the loss, albeit in a data-dependent way. We finally demonstrate the existence of a search phase for FastICA on ImageNet, and discuss how the strong non-Gaussianity of said images compensates for the poor sample complexity of FastICA.

Accurate spatial-temporal prediction of network-based travelers' requests is crucial for the effective policy design of ridesharing platforms. Having knowledge of the total demand between various locations in the upcoming time slots enables platforms to proactively prepare adequate supplies, thereby increasing the likelihood of fulfilling travelers' requests and redistributing idle drivers to areas with high potential demand to optimize the global supply-demand equilibrium. This paper delves into the prediction of Origin-Destination (OD) demands at a fine-grained spatial level, especially when confronted with an expansive set of local regions. While this task holds immense practical value, it remains relatively unexplored within the research community. To fill this gap, we introduce a novel prediction model called OD-CED, which comprises an unsupervised space coarsening technique to alleviate data sparsity and an encoder-decoder architecture to capture both semantic and geographic dependencies. Through practical experimentation, OD-CED has demonstrated remarkable results. It achieved an impressive reduction of up to 45% reduction in root-mean-square error and 60% in weighted mean absolute percentage error over traditional statistical methods when dealing with OD matrices exhibiting a sparsity exceeding 90%.