Feature transformation enhances downstream task performance by generating informative features through mathematical feature crossing. Despite the advancements in deep learning, feature transformation remains essential for structured data, where deep models often struggle to capture complex feature interactions. Prior literature on automated feature transformation has achieved success but often relies on heuristics or exhaustive searches, leading to inefficient and time-consuming processes. Recent works employ reinforcement learning (RL) to enhance traditional approaches through a more effective trial-and-error way. However, two limitations remain: 1) Dynamic feature expansion during the transformation process, which causes instability and increases the learning complexity for RL agents; 2) Insufficient cooperation and communication between agents, which results in suboptimal feature crossing operations and degraded model performance. To address them, we propose a novel heterogeneous multi-agent RL framework to enable cooperative and scalable feature transformation. The framework comprises three heterogeneous agents, grouped into two types, each designed to select essential features and operations for feature crossing. To enhance communication among these agents, we implement a shared critic mechanism that facilitates information exchange during feature transformation. To handle the dynamically expanding feature space, we tailor multi-head attention-based feature agents to select suitable features for feature crossing. Additionally, we introduce a state encoding technique during the optimization process to stabilize and enhance the learning dynamics of the RL agents, resulting in more robust and reliable transformation policies. Finally, we conduct extensive experiments to validate the effectiveness, efficiency, robustness, and interpretability of our model.

Reinforcement Learning (RL) traditionally relies on scalar reward signals, limiting its ability to leverage the rich semantic knowledge often available in real-world tasks. In contrast, humans learn efficiently by combining numerical feedback with language, prior knowledge, and common sense. We introduce Prompted Policy Search (ProPS), a novel RL method that unifies numerical and linguistic reasoning within a single framework. Unlike prior work that augment existing RL components with language, ProPS places a large language model (LLM) at the center of the policy optimization loop-directly proposing policy updates based on both reward feedback and natural language input. We show that LLMs can perform numerical optimization in-context, and that incorporating semantic signals, such as goals, domain knowledge, and strategy hints can lead to more informed exploration and sample-efficient learning. ProPS is evaluated across fifteen Gymnasium tasks, spanning classic control, Atari games, and MuJoCo environments, and compared to seven widely-adopted RL algorithms (e.g., PPO, SAC, TRPO). It outperforms all baselines on eight out of fifteen tasks and demonstrates substantial gains when provided with domain knowledge. These results highlight the potential of unifying semantics and numerics for transparent, generalizable, and human-aligned RL.

"Thinking with images" has emerged as an effective paradigm for advancing visual reasoning, extending beyond text-only chains of thought by injecting visual evidence into intermediate reasoning steps. However, existing methods fall short of human-like abstract visual thinking, as their flexibility is fundamentally limited by external tools. In this work, we introduce Monet, a training framework that enables multimodal large language models (MLLMs) to reason directly within the latent visual space by generating continuous embeddings that function as intermediate visual thoughts. We identify two core challenges in training MLLMs for latent visual reasoning: high computational cost in latent-vision alignment and insufficient supervision over latent embeddings, and address them with a three-stage distillation-based supervised fine-tuning (SFT) pipeline. We further reveal a limitation of applying GRPO to latent reasoning: it primarily enhances text-based reasoning rather than latent reasoning. To overcome this, we propose VLPO (Visual-latent Policy Optimization), a reinforcement learning method that explicitly incorporates latent embeddings into policy gradient updates. To support SFT, we construct Monet-SFT-125K, a high-quality text-image interleaved CoT dataset containing 125K real-world, chart, OCR, and geometry CoTs. Our model, Monet-7B, shows consistent gains across real-world perception and reasoning benchmarks and exhibits strong out-of-distribution generalization on challenging abstract visual reasoning tasks. We also empirically analyze the role of each training component and discuss our early unsuccessful attempts, providing insights for future developments in visual latent reasoning. Our model, data, and code are available at https://github.com/NOVAglow646/Monet.

While deep reinforcement learning agents demonstrate high performance across domains, their internal decision processes remain difficult to interp ret when evaluated only through performance metrics. In particular, it is poorly understoo d which input features agents rely on, how these dependencies evolve during training, and how t hey relate to behavior. We introduce a scientific methodology for analyzing the learni ng process through quantitative analysis of saliency. This approach aggregates saliency in formation at the object and modality level into hierarchical attention profiles, quantifyin g how agents allocate attention over time, thereby forming attention trajectories throughout t raining. Applied to Atari benchmarks, custom Pong environments, and muscle-actuated biom echanical user simulations in visuomotor interactive tasks, this methodology uncovers a lgorithm-specific attention biases, reveals unintended reward-driven strategies, and diagnos es overfitting to redundant sensory channels. These patterns correspond to measurable behavio ral differences, demonstrating empirical links between attention profiles, learning dynam ics, and agent behavior. To assess robustness of the attention profiles, we validate our finding s across multiple saliency methods and environments. The results establish attention traj ectories as a promising diagnostic axis for tracing how feature reliance develops during train ing and for identifying biases and vulnerabilities invisible to performance metrics alone.

Deductive reasoning is the process of deriving conclusions strictly from the given premises, without relying on external knowledge. We define honesty in this setting as a model's ability to respond only when the conclusion is logically entailed by the premises, and to abstain otherwise. However, current language models often fail to reason honestly, producing unwarranted answers when the input is insufficient. To study this challenge, we formulate honest deductive reasoning as multi-step tasks where models must either derive the correct conclusion or abstain. We curate two datasets from graph structures, one for linear algebra and one for logical inference, and introduce unanswerable cases by randomly perturbing an edge in half of the instances. We find that prompting and existing training methods, including GRPO with or without supervised fine-tuning initialization, struggle on these tasks. In particular, GRPO optimize only for final task outcomes, leaving models vulnerable to collapse when negative rewards dominate early training. To address this, we propose ACNCHOR, a reinforcement learning method that injects ground truth trajectories into rollouts, preventing early training collapse. Our results demonstrate that this method stabilizes learning and significantly improves the overall reasoning performance, underscoring the importance of training dynamics for enabling honest deductive reasoning in language models.

Unsupervised skill discovery in reinforcement learning (RL) aims to learn diverse behaviors without relying on external rewards. However, current methods often overlook the periodic nature of learned skills, focusing instead on increasing the mutual dependence between states and skills or maximizing the distance traveled in latent space. Considering that many robotic tasks - particularly those involving locomotion - require periodic behaviors across varying timescales, the ability to discover diverse periodic skills is essential. Motivated by this, we propose Periodic Skill Discovery (PSD), a framework that discovers periodic behaviors in an unsupervised manner. The key idea of PSD is to train an encoder that maps states to a circular latent space, thereby naturally encoding periodicity in the latent representation. By capturing temporal distance, PSD can effectively learn skills with diverse periods in complex robotic tasks, even with pixel-based observations. We further show that these learned skills achieve high performance on downstream tasks such as hurdling. Moreover, integrating PSD with an existing skill discovery method offers more diverse behaviors, thus broadening the agent's repertoire. Our code and demos are available at https://jonghaepark.github.io/psd/

To enable embodied agents to operate effectively over extended timeframes, it is crucial to develop models that form and access memories to stay contextualized in their environment. In the current paradigm of training transformer-based policies for embodied sequential decision-making tasks, visual inputs often overwhelm the context limits of transformers, while humans can maintain and utilize a lifetime of experience compressed as memories. Significant compression is possible in principle, as much of the input is irrelevant and can be abstracted. However, existing approaches predominantly focus on either recurrent models with fixed-size memory or transformers with full-context reliance. In this work, we propose Memo, a transformer-based architecture and training recipe for reinforcement learning (RL) on memory-intensive, long-horizon tasks. Memo incorporates the creation and retrieval of memory by interleaving periodic summarization tokens with the inputs of a model during training. We demonstrate Memo's effectiveness on a gridworld meta-RL benchmark and a multi-object navigation task in photo-realistic indoor settings. Memo outperforms naive long-context transformer baselines while being more compute and storage efficient. Additionally, Memo generalizes better to longer contexts at inference time and remains robust in streaming settings, where historical context must be truncated to fit inference constraints. Our code is available at: https://github.com/gunshi/memo.

Abstract-- Robotic automation is accelerating scientific discovery by reducing manual effort in laboratory workflows. However, precise manipulation of powders remains challenging, particularly in tasks such as transport that demand accuracy and stability. We propose a trajectory optimisation framework for powder transport in laboratory settings, which integrates differentiable physics simulation for accurate modelling of granular dynamics, low-dimensional skill-space parameterisation to reduce optimisation complexity, and a curriculum-based strategy that progressively refines task competence over long horizons. This formulation enables end-to-end optimisation of contact-rich robot trajectories while maintaining stability and convergence efficiency. Experimental results demonstrate that the proposed method achieves superior task success rates and stability compared to the reinforcement learning baseline.

Rapid post-disaster road damage assessment is critical for effective emergency response, yet traditional optimization methods suffer from excessive computational time and require domain knowledge for algorithm design, making them unsuitable for time-sensitive disaster scenarios. This study proposes an attention-based encoder-decoder model (AEDM) for rapid drone routing decision in post-disaster road damage assessment. The method employs deep reinforcement learning to determine high-quality drone assessment routes without requiring algorithmic design knowledge. A network transformation method is developed to convert link-based routing problems into equivalent node-based formulations, while a synthetic road network generation technique addresses the scarcity of large-scale training datasets. The model is trained using policy optimization with multiple optima (POMO) with multi-task learning capabilities to handle diverse parameter combinations. Experimental results demonstrate two key strengths of AEDM: it outperforms commercial solvers by 20--71\% and traditional heuristics by 23--35\% in solution quality, while achieving rapid inference (1--2 seconds) versus 100--2,000 seconds for traditional methods. The model exhibits strong generalization across varying problem scales, drone numbers, and time constraints, consistently outperforming baseline methods on unseen parameter distributions and real-world road networks. The proposed method effectively balances computational efficiency with solution quality, making it particularly suitable for time-critical disaster response applications where rapid decision-making is essential for saving lives. The source code for AEDM is publicly available at https://github.com/PJ-HTU/AEDM-for-Post-disaster-road-assessment.

Reinforcement learning from human feedback (RLHF) has emerged as a key enabling technology for aligning AI behaviour with human preferences. The traditional way to collect data in RLHF is via pairwise comparisons: human raters are asked to indicate which one of two samples they prefer. We present an interactive visualisation that better exploits the human visual ability to compare and explore whole groups of samples. The interface is comprised of two linked views: 1) an exploration view showing a contextual overview of all sampled behaviours organised in a hierarchical clustering structure; and 2) a comparison view displaying two selected groups of behaviours for user queries. Users can efficiently explore large sets of behaviours by iterating between these two views. Additionally, we devised an active learning approach suggesting groups for comparison. As shown by our evaluation in six simulated robotics tasks, our approach increases the final rewards by 69.34%. It leads to lower error rates and better policies. We open-source the code that can be easily integrated into the RLHF training loop, supporting research on human-AI alignment.