World models learn to simulate environment dynamics from experience, enabling sample-efficient reinforcement learning. But what do these models actually represent internally? We apply interpretability techniques–including linear and nonlinear probing, causal interventions, and attention analysis–to two architecturally distinct world models: IRIS (discrete token transformer) and DIAMOND (continuous diffusion UNet), trained on Atari Breakout and Pong. Using linear probes, we find that both models develop linearly decodable representations of game state variables (object positions, scores), with MLP probes yielding only marginally higher R^2, confirming that these representations are approximately linear. Causal interventions–shifting hidden states along probe-derived directions–produce correlated changes in model predictions, providing evidence that representations are functionally used rather than merely correlated. Analysis of IRIS attention heads reveals spatial specialization: specific heads attend preferentially to tokens overlapping with game objects. Multi-baseline token ablation experiments consistently identify object-containing tokens as disproportionately important. Our findings provide interpretability evidence that learned world models develop structured, approximately linear internal representations of environment state across two games and two architectures.

Learning to bid in repeated multi-unit auctions with bandit feedback poses a fundamental challenge. Existing methods often rely on rigid explore-then-exploit schedules, assume stationary adversaries, and optimize solely for bidder utility, thereby limiting adaptability and strategic robustness. To address these limitations, we introduce the A3M framework, which integrates adaptive deep reinforcement learning (DRL), explicit adversarial reasoning, and principled multi-objective reward design for online auction strategy optimization. A3M employs an actor-critic DRL backbone to dynamically balance exploration and exploitation, an opponent model for fictitious play against non-stationary adversaries, and a composite reward function to jointly maximize utility, auctioneer revenue, and fairness. We provide the first comprehensive empirical evaluation of this integrated approach against established baselines in both discriminatory and uniform price auctions. Results show that A3M reduces final regret by 30–40% in standard settings, maintains robust performance against adversarial strategy shifts, scales favorably with the number of units K, and enables tunable multi-objective trade-offs. An extensive ablation study confirms the necessity of each core component. Our work establishes A3M as a powerful and flexible framework for learning in complex auction environments.

Self-improvement training, where models learn from self-generated solutions, promises sustained capability gains but suffers from a pervasive failure mode: across multiple rounds, compounding reasoning errors cause accuracy to stall or degrade. We trace this drift to standard filtering criteria that retain solutions based solely on final answer correctness, which lets lucky guesses (correct answers with flawed reasoning) contaminate the training data. We propose Verified Self-Improvement (VSI), a framework that conditions data retention on step-level structural integrity rather than just the final output. VSI validates solutions by recomputing arithmetic steps via a computer-algebra library (sympy), checking intermediate consistency, and enforcing domain constraints. Evaluating VSI on GSM8K with Qwen3-4B-Thinking across 5 rounds of self-improvement against four baselines (no verification, outcome verification, majority voting, and VSI with DPO) shows that VSI rejects approximately 34

Time-varying non-stationary channels, with complex dynamic variations and temporal evolution characteristics, have significant challenges in channel modeling and communication system performance evaluation. Most existing methods of time-varying channel modeling focus on predicting channel state at a given moment or simulating short-term channel fluctuations, which are unable to capture the long-term evolution of the channel. This paper emphasizes the generation of long-term dynamic channel to fully capture evolution of non-stationary channel properties. The generated channel not only reflects temporal dynamics but also ensures consistent stationarity. We propose a hybrid deep learning framework that combines conditional generative adversarial networks (CGAN) with long short-term memory (LSTM) networks. A stationarity-constrained approach is designed to ensure temporal correlation of the generated time-series channel. This method can generate channel with required temporal non-stationarity. The model is validated by comparing channel statistical features, and the results show that the generated channel is in good agreement with raw channel and provides good performance in terms of non-stationarity.

The Structured Dialogue System, referred to as SuDoSys, is an innovative Large Language Model (LLM)-based chatbot designed to provide psychological counseling. SuDoSys leverages the World Health Organization (WHO)'s Problem Management Plus (PM+) guidelines to deliver stage-aware multi-turn dialogues. Existing methods for employing an LLM in multi-turn psychological counseling typically involve direct fine-tuning using generated dialogues, often neglecting the dynamic stage shifts of counseling sessions. Unlike previous approaches, SuDoSys considers the different stages of counseling and stores essential information throughout the counseling process, ensuring coherent and directed conversations. The system employs an LLM, a stage-aware instruction generator, a response unpacker, a topic database, and a stage controller to maintain dialogue flow. In addition, we propose a novel technique that simulates counseling clients to interact with the evaluated system and evaluate its performance automatically. When assessed using both objective and subjective evaluations, SuDoSys demonstrates its effectiveness in generating logically coherent responses. The system's code and program scripts for evaluation are open-sourced.

In this paper, we present a terrain modeling approach based on mid-air sketching. Users can outline the key spatial features of the terrain with the assistance of sparse 3D strokes. Since the 3D sketch has inherent structural advantages in depicting the spatial and geometric attributes of volumetric shapes, users can create personalized terrains. Experiments demonstrate that the proposed approach has significant advantages in creative freedom and artistic control in some scenarios. Finally, we hope that the approach in this paper can provide a new idea for the automatic generation of 3D content.

The learnware paradigm offers a novel approach to machine learning by enabling users to reuse a set of well-trained models for tasks beyond the models' original purposes. It eliminates the need to build models from scratch, instead relying on specifications (representations of a model's capabilities) to identify and leverage the most suitable models for new tasks. While learnware has proven effective in many scenarios, its application to language models has remained largely unexplored. At the same time, large language models (LLMs) have demonstrated remarkable universal question-answering abilities, yet they face challenges in specialized scenarios due to data scarcity, privacy concerns, and high computational costs, thus more and more specialized small language models (SLMs) are being trained for specific domains. To address these limitations systematically, the learnware paradigm provides a promising solution by enabling maximum utilization of specialized SLMs, and allowing users to identify and reuse them in a collaborative and privacy-preserving manner. This paper presents a preliminary attempt to apply the learnware paradigm to language models. We simulated a learnware system comprising approximately 100 learnwares of specialized SLMs with 8B parameters, fine-tuned across finance, healthcare, and mathematics domains. Each learnware contains an SLM and a specification, which enables users to identify the most relevant models without exposing their own data. Experimental results demonstrate promising performance: by selecting one suitable learnware for each task-specific inference, the system outperforms the base SLMs on all benchmarks. Compared to LLMs, the system outperforms Qwen1.5-110B, Qwen2.5-72B, and Llama3.1-70B-Instruct by at least 14 Flan-PaLM-540B (ranked 7th on the Open Medical LLM Leaderboard) in medical domain tasks.

Terrains are key elements in 3D scene creation for applications such as VR/AR, simulation, and games. Current sketch-based terrain modeling tools rely on backtracking workflows where users usually repeatedly undo and revise previous strokes to adjust the terrain model. This backtracking tends to interrupt creative flow and increase cognitive effort. We present ForwardTerrain, an interactive system that enables one-pass terrain modeling through mid-air sketching without backtracking. Instead of reverting to earlier sketches, users can directly edit and extend the existing terrain model, supporting a continuous and fluid authoring experience. Powered by a StyleGAN2-based generator, our system interprets sketches into plausible terrain shapes in real time. In a controlled user study (N = 24), ForwardTerrain significantly improved both modeling efficiency and accuracy compared to backtracking workflows, reducing task time by 15% and 11%, and increasing accuracy by 13% and 20% in generation and modification tasks, respectively. Participants also reported higher perceived creativity support and lower cognitive load, particularly among non-expert users. These results highlight the value of one-pass workflows in 3D modeling, fostering smoother creative experiences and greater accessibility.

Developing policies that can adjust to non-stationary environments is essential for real-world reinforcement learning applications. However, learning such adaptable policies in offline settings, with only a limited set of pre-collected trajectories, presents significant challenges. A key difficulty arises because the limited offline data makes it hard for the context encoder to differentiate between changes in the environment dynamics and shifts in the behavior policy, often leading to context misassociations. To address this issue, we introduce a novel approach called Debiased Offline Representation for fast online Adaptation (DORA). DORA incorporates an information bottleneck principle that maximizes mutual information between the dynamics encoding and the environmental data, while minimizing mutual information between the dynamics encoding and the actions of the behavior policy. We present a practical implementation of DORA, leveraging tractable bounds of the information bottleneck principle. Our experimental evaluation across six benchmark MuJoCo tasks with variable parameters demonstrates that DORA not only achieves a more precise dynamics encoding but also significantly outperforms existing baselines in terms of performance.

Alphas are pivotal in providing signals for quantitative trading. The industry highly values the discovery of formulaic alphas for their interpretability and ease of analysis, compared with the expressive yet overfitting-prone black-box alphas. In this work, we focus on discovering formulaic alphas. Prior studies on automatically generating a collection of formulaic alphas were mostly based on genetic programming (GP), which is known to suffer from the problems of being sensitive to the initial population, converting to local optima, and slow computation speed. Recent efforts employing deep reinforcement learning (DRL) for alpha discovery have not fully addressed key practical considerations such as alpha correlations and validity, which are crucial for their effectiveness. In this work, we propose a novel framework for alpha discovery using DRL by formulating the alpha discovery process as program construction. Our agent, $\text{Alpha}^2$, assembles an alpha program optimized for an evaluation metric. A search algorithm guided by DRL navigates through the search space based on value estimates for potential alpha outcomes. The evaluation metric encourages both the performance and the diversity of alphas for a better final trading strategy. Our formulation of searching alphas also brings the advantage of pre-calculation dimensional analysis, ensuring the logical soundness of alphas, and pruning the vast search space to a large extent. Empirical experiments on real-world stock markets demonstrates $\text{Alpha}^2$'s capability to identify a diverse set of logical and effective alphas, which significantly improves the performance of the final trading strategy. The code of our method is available at https://github.com/x35f/alpha2.

Student simulation in online education is important to address dynamic learning behaviors of students with diverse backgrounds. Existing simulation models based on deep learning usually need massive training data, lacking prior knowledge in educational contexts. Large language models (LLMs) may contain such prior knowledge since they are pre-trained from a large corpus. However, because student behaviors are dynamic and multifaceted with individual differences, directly prompting LLMs is not robust nor accurate enough to capture fine-grained interactions among diverse student personas, learning behaviors, and learning outcomes. This work tackles this problem by presenting a newly annotated fine-grained large-scale dataset and proposing EduAgent, a novel generative agent framework incorporating cognitive prior knowledge (i.e., theoretical findings revealed in cognitive science) to guide LLMs to first reason correlations among various behaviors and then make simulations. Our two experiments show that EduAgent could not only mimic and predict learning behaviors of real students but also generate realistic learning behaviors of virtual students without real data.

A learning task is when students finish a certain learning activity or reach a particular learning objective in a set amount of time. According to the demands of the new curriculum for 2022, it is inevitable that learning task designs will incorporate progressive learning into the teaching process. Task group designs will also be used to help students achieve some degree of all-around development through progressive learning. This essay explores the teacher s new route in the classroom by using Xue Fagan s Wang Rong didn t pick roadside plums lesson as an example and the learning theory of thinking.

Human visual attention is susceptible to social influences. In education, peer effects impact student learning, but their precise role in modulating attention remains unclear. Our experiment (N=311) demonstrates that displaying peer visual attention regions when students watch online course videos enhances their focus and engagement. However, students retain adaptability in following peer attention cues. Overall, guided peer attention improves learning experiences and outcomes. These findings elucidate how peer visual attention shapes students' gaze patterns, deepening understanding of peer influence on learning. They also offer insights into designing adaptive online learning interventions leveraging peer attention modelling to optimize student attentiveness and success.

Student simulation presents a transformative approach to enhance learning outcomes, advance educational research, and ultimately shape the future of effective pedagogy. We explore the feasibility of using large language models (LLMs), a remarkable achievement in AI, to simulate student learning behaviors. Unlike conventional machine learning based prediction, we leverage LLMs to instantiate virtual students with specific demographics and uncover intricate correlations among learning experiences, course materials, understanding levels, and engagement. Our objective is not merely to predict learning outcomes but to replicate learning behaviors and patterns of real students. We validate this hypothesis through three experiments. The first experiment, based on a dataset of N = 145, simulates student learning outcomes from demographic data, revealing parallels with actual students concerning various demographic factors. The second experiment (N = 4524) results in increasingly realistic simulated behaviors with more assessment history for virtual students modelling. The third experiment (N = 27), incorporating prior knowledge and course interactions, indicates a strong link between virtual students' learning behaviors and fine-grained mappings from test questions, course materials, engagement and understanding levels. Collectively, these findings deepen our understanding of LLMs and demonstrate its viability for student simulation, empowering more adaptable curricula design to enhance inclusivity and educational effectiveness.

Imitation learning aims to mimic the behavior of experts without explicit reward signals. Passive imitation learning methods which use static expert datasets typically suffer from compounding error, low sample efficiency, and high hyper-parameter sensitivity. In contrast, active imitation learning methods solicit expert interventions to address the limitations. However, recent active imitation learning methods are designed based on human intuitions or empirical experience without theoretical guarantee. In this paper, we propose a novel active imitation learning framework based on a teacher-student interaction model, in which the teacher's goal is to identify the best teaching behavior and actively affect the student's learning process. By solving the optimization objective of this framework, we propose a practical implementation, naming it AdapMen. Theoretical analysis shows that AdapMen can improve the error bound and avoid compounding error under mild conditions. Experiments on the MetaDrive benchmark and Atari 2600 games validate our theoretical analysis and show that our method achieves near-expert performance with much less expert involvement and total sampling steps than previous methods. The code is available at https://github.com/liuxhym/AdapMen.

In view of the practical application of target detection and recognition tasks for remote sensing, an end-to-end aircraft target detection and fine-grained recognition framework is proposed. It can accurately and quickly implement detection and recognition in an end-to-end way. The main network of the framework adopts the design ideas of target detection and fine-grained recognition methods using candidate region extract and visual attention, making sure the accuracy of detection and recognition. Then, to solve the problem of high false detection rate and missed detection rate of densely arranged targets, we propose the re-detection mechanism. To minimize the large amounts of calculations of deep networks and improve real-time performance, we introduce depthwise separable convolution to optimize networks. Finally, a weight mapping idea based on transfer learning is adopted to solve the problem of data labeling and also helps the detection and fine-grained recognition. The results prove that the proposed framework has good robustness, versatility, and efficiency in aircraft detection and fine-grained recognition tasks.