Radiology report generation is critical for efficiency but current models lack the structured reasoning of experts, hindering clinical trust and explainability by failing to link visual findings to precise anatomical locations. This paper introduces BoxMed-RL, a groundbreaking unified training framework for generating spatially verifiable and explainable radiology reports. Built on a large vision-language model, BoxMed-RL revolutionizes report generation through two integrated phases: (1) In the Pretraining Phase, we refine the model via medical concept learning, using Chain-of-Thought supervision to internalize the radiologist-like workflow, followed by spatially verifiable reinforcement, which applies reinforcement learning to align medical findings with bounding boxes. (2) In the Downstream Adapter Phase, we freeze the pretrained weights and train a downstream adapter to ensure fluent and clinically credible reports. This framework precisely mimics radiologists' workflow, compelling the model to connect high-level medical concepts with definitive anatomical evidence. Extensive experiments on public datasets demonstrate that BoxMed-RL achieves an average 7% improvement in both METEOR and ROUGE-L metrics compared to state-of-the-art methods. An average 5% improvement in large language model-based metrics further underscores BoxMed-RL's robustness in generating high-quality radiology reports.

The integration of unmanned aerial vehicles (UAVs) into cellular networks presents significant mobility management challenges, primarily due to frequent handovers caused by probabilistic line-of-sight conditions with multiple ground base stations (BSs). To tackle these challenges, reinforcement learning (RL)-based methods, particularly deep Q-networks (DQN), have been employed to optimize handover decisions dynamically. However, a major drawback of these learning-based approaches is their black-box nature, which limits interpretability in the decision-making process. This paper introduces an explainable AI (XAI) framework that incorporates Shapley Additive Explanations (SHAP) to provide deeper insights into how various state parameters influence handover decisions in a DQN-based mobility management system. By quantifying the impact of key features such as reference signal received power (RSRP), reference signal received quality (RSRQ), buffer status, and UAV position, our approach enhances the interpretability and reliability of RL-based handover solutions. To validate and compare our framework, we utilize real-world network performance data collected from UAV flight trials. Simulation results show that our method provides intuitive explanations for policy decisions, effectively bridging the gap between AI-driven models and human decision-makers.

This paper addresses the challenge of navigation in large, visually complex environments with sparse rewards. We propose a method that uses object-oriented macro actions grounded in a topological map, allowing a simple Deep Q-Network (DQN) to learn effective navigation policies. The agent builds a map by detecting objects from RGBD input and selecting discrete macro actions that correspond to navigating to these objects. This abstraction drastically reduces the complexity of the underlying reinforcement learning problem and enables generalization to unseen environments. We evaluate our approach in a photorealistic 3D simulation and show that it significantly outperforms a random baseline under both immediate and terminal reward conditions. Our results demonstrate that topological structure and macro-level abstraction can enable sample-efficient learning even from pixel data.

Event-based cameras are bio-inspired sensors that detect light changes asynchronously for each pixel. They are increasingly used in fields like computer vision and robotics because of several advantages over traditional frame-based cameras, such as high temporal resolution, low latency, and high dynamic range. As with any camera, the output's quality depends on how well the camera's settings, called biases for event-based cameras, are configured. While frame-based cameras have advanced automatic configuration algorithms, there are very few such tools for tuning these biases. A systematic testing framework would require observing the same scene with different biases, which is tricky since event cameras only generate events when there is movement. Event simulators exist, but since biases heavily depend on the electrical circuit and the pixel design, available simulators are not well suited for bias tuning. To allow reproducibility, we present BiasBench, a novel event dataset containing multiple scenes with settings sampled in a grid-like pattern. We present three different scenes, each with a quality metric of the downstream application. Additionally, we present a novel, RL-based method to facilitate online bias adjustments.

Accepted to the Generative Models for Robot Learning Workshop at ICLR 2025O FFLINEL EARNING OF C ONTROLLABLED IVERSE B E-HAVIORS Mathieu Petitbois *,1, R emy Portelas 1, Sylvain Lamprier 2, Ludovic Denoyer 3 1 Ubisoft La Forge 2 University of Angers 3 H Company A BSTRACT Imitation Learning (IL) techniques aim to replicate human behaviors in specific tasks. While IL has gained prominence due to its effectiveness and efficiency, traditional methods often focus on datasets collected from experts to produce a single efficient policy. Recently, extensions have been proposed to handle datasets of diverse behaviors by mainly focusing on learning transition-level diverse policies or on performing entropy maximization at the trajectory level. While these methods may lead to diverse behaviors, they may not be sufficient to reproduce the actual diversity of demonstrations or to allow controlled trajectory generation. To overcome these drawbacks, we propose a different method based on two key features: a) Temporal Consistency that ensures consistent behaviors across entire episodes and not just at the transition level as well as b) Controllability obtained by constructing a latent space of behaviors that allows users to selectively activate specific behaviors based on their requirements. We compare our approach to state-of-the-art methods over a diverse set of tasks and environments. For robotics, learning from human experts allows to reach human-level performance without any controller hard coding or expensive interaction with simulated or real environments.

This paper introduces an approach for developing surrogate environments in reinforcement learning (RL) using the Sparse Identification of Nonlinear Dynamics (SINDy) algorithm. We demonstrate the effectiveness of our approach through extensive experiments in OpenAI Gym environments, particularly Mountain Car and Lunar Lander. Our results show that SINDy-based surrogate models can accurately capture the underlying dynamics of these environments while reducing computational costs by 20-35%. With only 75 interactions for Mountain Car and 1000 for Lunar Lander, we achieve state-wise correlations exceeding 0.997, with mean squared errors as low as 3.11e-06 for Mountain Car velocity and 1.42e-06 for LunarLander position. RL agents trained in these surrogate environments require fewer total steps (65,075 vs. 100,000 for Mountain Car and 801,000 vs. 1,000,000 for Lunar Lander) while achieving comparable performance to those trained in the original environments, exhibiting similar convergence patterns and final performance metrics. This work contributes to the field of model-based RL by providing an efficient method for generating accurate, interpretable surrogate environments.

This work presents reinforcement learning (RL)-driven data augmentation to improve the generalization of vision-action (VA) models for dexterous grasping. While real-to-sim-to-real frameworks, where a few real demonstrations seed large-scale simulated data, have proven effective for VA models, applying them to dexterous settings remains challenging: obtaining stable multi-finger contacts is nontrivial across diverse object shapes. To address this, we leverage RL to generate contact-rich grasping data across varied geometries. In line with the real-to-sim-to-real paradigm, the grasp skill is formulated as a parameterized and tunable reference trajectory refined by a residual policy learned via RL. This modular design enables trajectory-level control that is both consistent with real demonstrations and adaptable to diverse object geometries. A vision-conditioned policy trained on simulation-augmented data demonstrates strong generalization to unseen objects, highlighting the potential of our approach to alleviate the data bottleneck in training VA models.

Auto-bidding, with its strong capability to optimize bidding decisions within dynamic and competitive online environments, has become a pivotal strategy for advertising platforms. Existing approaches typically employ rule-based strategies or Reinforcement Learning (RL) techniques. However, rule-based strategies lack the flexibility to adapt to time-varying market conditions, and RL-based methods struggle to capture essential historical dependencies and observations within Markov Decision Process (MDP) frameworks. Furthermore, these approaches often face challenges in ensuring strategy adaptability across diverse advertising objectives. Additionally, as offline training methods are increasingly adopted to facilitate the deployment and maintenance of stable online strategies, the issues of documented behavioral patterns and behavioral collapse resulting from training on fixed offline datasets become increasingly significant. To address these limitations, this paper introduces a novel offline Generative Auto-bidding framework with Value-Guided Explorations (GAVE). GAVE accommodates various advertising objectives through a score-based Return-To-Go (RTG) module. Moreover, GAVE integrates an action exploration mechanism with an RTG-based evaluation method to explore novel actions while ensuring stability-preserving updates. A learnable value function is also designed to guide the direction of action exploration and mitigate Out-of-Distribution (OOD) problems. Experimental results on two offline datasets and real-world deployments demonstrate that GAVE outperforms state-of-the-art baselines in both offline evaluations and online A/B tests. By applying the core methods of this framework, we proudly secured first place in the NeurIPS 2024 competition, 'AIGB Track: Learning Auto-Bidding Agents with Generative Models'.

Generative information retrieval (GenIR) is a promising neural retrieval paradigm that formulates document retrieval as a document identifier (docid) generation task, allowing for end-to-end optimization toward a unified global retrieval objective. However, existing GenIR models suffer from token-level misalignment, where models trained to predict the next token often fail to capture document-level relevance effectively. While reinforcement learning-based methods, such as reinforcement learning from relevance feedback (RLRF), aim to address this misalignment through reward modeling, they introduce significant complexity, requiring the optimization of an auxiliary reward function followed by reinforcement fine-tuning, which is computationally expensive and often unstable. To address these challenges, we propose direct document relevance optimization (DDRO), which aligns token-level docid generation with document-level relevance estimation through direct optimization via pairwise ranking, eliminating the need for explicit reward modeling and reinforcement learning. Experimental results on benchmark datasets, including MS MARCO document and Natural Questions, show that DDRO outperforms reinforcement learning-based methods, achieving a 7.4% improvement in MRR@10 for MS MARCO and a 19.9% improvement for Natural Questions. These findings highlight DDRO's potential to enhance retrieval effectiveness with a simplified optimization approach. By framing alignment as a direct optimization problem, DDRO simplifies the ranking optimization pipeline of GenIR models while offering a viable alternative to reinforcement learning-based methods.

Do We Need Transformers to Play FPS Video Games? Abstract --In this paper, we explore the Transformer based architectures for reinforcement learning in both online and offline settings within the Doom game environment. Our investigation focuses on two primary approaches: Deep Transformer Q-learning Networks (DTQN) for online learning [1] and Decision Transformers (DT) for offline reinforcement learning [2]. DTQN leverages the sequential modelling capabilities of Transformers to enhance Q-learning in partially observable environments,while Decision Transformers repurpose sequence modelling techniques to enable offline agents to learn from past trajectories without direct interaction with the environment. We conclude that while Transformers might have performed well in Atari games, more traditional methods perform better than Transformer based method in both the settings in the VizDoom environment [3] Q-networks traditionally have been used in various Atari game environments [4] and in partially observable environments like in the Doom 1993 game [5].