arXiv subjects

Ruiyu Wang

Publications and source records attributed to Ruiyu Wang.

34 records · Page 2Linked to original sources

arXiv2025

Feature Extractor or Decision Maker: Rethinking the Role of Visual Encoders in Visuomotor Policies

An end-to-end (E2E) visuomotor policy is typically treated as a unified whole, but recent approaches using out-of-domain (OOD) data to pretrain the visual encoder have cleanly separated the visual encoder from the network, with the remainder referred to as the policy. We propose Visual Alignment Testing, an experimental framework designed to evaluate the validity of this functional separation. Our results indicate that in E2E-trained models, visual encoders actively contribute to decision-making resulting from motor data supervision, contradicting the assumed functional separation. In contrast, OOD-pretrained models, where encoders lack this capability, experience an average performance drop of 42\% in our benchmark results, compared to the state-of-the-art performance achieved by E2E policies. We believe this initial exploration of visual encoders' role can provide a first step towards guiding future pretraining methods to address their decision-making ability, such as developing task-conditioned or context-aware encoders.

cs.RO↗

arXiv2025

Event-aware analysis of cross-city visitor flows using large language models and social media data

Public events, such as music concerts and fireworks displays, can cause irregular surges in cross-city travel demand, leading to potential overcrowding, travel delays, and public safety concerns. To better anticipate and accommodate such demand surges, it is essential to estimate cross-city visitor flows with awareness of public events. Although prior studies typically focused on the effects of a single mega event or disruptions around a single venue, this study introduces a generalizable framework to analyze visitor flows under diverse and concurrent events. We propose to leverage large language models (LLMs) to extract event features from multi-source online information and massive user-generated content on social media platforms. Specifically, social media popularity metrics are designed to capture the effects of online promotion and word-of-mouth in attracting visitors. An event-aware machine learning model is then adopted to uncover the specific impacts of different event features and ultimately predict visitor flows for upcoming events. Using Hong Kong as a case study, the framework is applied to predict daily flows of mainland Chinese visitors arriving at the city, achieving a testing R-squared of over 85%. We further investigate the heterogeneous event impacts on visitor numbers across different event types and major travel modes. Both promotional popularity and word-of-mouth popularity are found to be associated with increased visitor flows, but the specific effects vary by the event type. This association is more pronounced among visitors arriving by metro and high-speed rail, while it has less effect on air travelers. The findings can facilitate coordinated measures across government agencies and guide specialized transport policies, such as shuttle transit services to event venues, and comprehensive on-site traffic management strategies.

cs.SI↗

arXiv2025

Exponentially Tilted Thermodynamic Maps (expTM): Predicting Phase Transitions Across Temperature, Pressure, and Chemical Potential

Predicting and characterizing phase transitions is crucial for understanding generic physical phenomena such as crystallization, protein folding and others. However, directly observing phase transitions is not always easy, and often one has limited observations far from the phase boundary and measured under some specific thermodynamic conditions. In this study, we propose a statistical physics and Generative AI driven framework that can take such limited information to generate samples of different phases under arbitrary thermodynamic conditions, which we name Exponentially Tilted Thermodynamic Maps (expTM). The central idea is to map collected data into a tractable simple prior expressed as an exponentially tilted Gaussian. We demonstrate how the variance and mean of the prior can be correlated with pairs of thermodynamic control variables, including temperature, pressure, and chemical potential. This gives us the ability to generate thermodynamically correct samples under any values of the control variables. To demonstrate the practical applicability of this approach, we use expTM to sample the lattice gas models with the Grand Canonical ensemble, capturing phase transitions under varying chemical potentials and temperatures. We further demonstrate how expTM can model the isothermal-isobaric ensemble, with which we predict different phases of CO2 under varying pressure conditions. Both examples are trained on very limited data far from the phase boundary. These results establish expTM as a robust tool for understanding phase transitions across diverse thermodynamic conditions requiring only a small number of observations.

cond-mat.stat-mech↗

arXiv2025

Lifelong Learning with Task-Specific Adaptation: Addressing the Stability-Plasticity Dilemma

Lifelong learning (LL) aims to continuously acquire new knowledge while retaining previously learned knowledge. A central challenge in LL is the stability-plasticity dilemma, which requires models to balance the preservation of previous knowledge (stability) with the ability to learn new tasks (plasticity). While parameter-efficient fine-tuning (PEFT) has been widely adopted in large language models, its application to lifelong learning remains underexplored. To bridge this gap, this paper proposes AdaLL, an adapter-based framework designed to address the dilemma through a simple, universal, and effective strategy. AdaLL co-trains the backbone network and adapters under regularization constraints, enabling the backbone to capture task-invariant features while allowing the adapters to specialize in task-specific information. Unlike methods that freeze the backbone network, AdaLL incrementally enhances the backbone's capabilities across tasks while minimizing interference through backbone regularization. This architectural design significantly improves both stability and plasticity, effectively eliminating the stability-plasticity dilemma. Extensive experiments demonstrate that AdaLL consistently outperforms existing methods across various configurations, including dataset choices, task sequences, and task scales.

cs.LG↗

arXiv2025

RealCraft: Attention Control as A Tool for Zero-Shot Consistent Video Editing

Even though large-scale text-to-image generative models show promising performance in synthesizing high-quality images, applying these models directly to image editing remains a significant challenge. This challenge is further amplified in video editing due to the additional dimension of time. This is especially the case for editing real-world videos as it necessitates maintaining a stable structural layout across frames while executing localized edits without disrupting the existing content. In this paper, we propose RealCraft, an attention-control-based method for zero-shot real-world video editing. By swapping cross-attention for new feature injection and relaxing spatial-temporal attention of the editing object, we achieve localized shape-wise edit along with enhanced temporal consistency. Our model directly uses Stable Diffusion and operates without the need for additional information. We showcase the proposed zero-shot attention-control-based method across a range of videos, demonstrating shape-wise, time-consistent and parameter-free editing in videos of up to 64 frames.

cs.CV↗

arXiv2024

PACA: Perspective-Aware Cross-Attention Representation for Zero-Shot Scene Rearrangement

Scene rearrangement, like table tidying, is a challenging task in robotic manipulation due to the complexity of predicting diverse object arrangements. Web-scale trained generative models such as Stable Diffusion can aid by generating natural scenes as goals. To facilitate robot execution, object-level representations must be extracted to match the real scenes with the generated goals and to calculate object pose transformations. Current methods typically use a multi-step design that involves separate models for generation, segmentation, and feature encoding, which can lead to a low success rate due to error accumulation. Furthermore, they lack control over the viewing perspectives of the generated goals, restricting the tasks to 3-DoF settings. In this paper, we propose PACA, a zero-shot pipeline for scene rearrangement that leverages perspective-aware cross-attention representation derived from Stable Diffusion. Specifically, we develop a representation that integrates generation, segmentation, and feature encoding into a single step to produce object-level representations. Additionally, we introduce perspective control, thus enabling the matching of 6-DoF camera views and extending past approaches that were limited to 3-DoF top-down views. The efficacy of our method is demonstrated through its zero-shot performance in real robot experiments across various scenes, achieving an average matching accuracy and execution success rate of 87% and 67%, respectively.

cs.RO↗

arXiv2024

Double Oracle Neural Architecture Search for Game Theoretic Deep Learning Models

In this paper, we propose a new approach to train deep learning models using game theory concepts including Generative Adversarial Networks (GANs) and Adversarial Training (AT) where we deploy a double-oracle framework using best response oracles. GAN is essentially a two-player zero-sum game between the generator and the discriminator. The same concept can be applied to AT with attacker and classifier as players. Training these models is challenging as a pure Nash equilibrium may not exist and even finding the mixed Nash equilibrium is difficult as training algorithms for both GAN and AT have a large-scale strategy space. Extending our preliminary model DO-GAN, we propose the methods to apply the double oracle framework concept to Adversarial Neural Architecture Search (NAS for GAN) and Adversarial Training (NAS for AT) algorithms. We first generalize the players' strategies as the trained models of generator and discriminator from the best response oracles. We then compute the meta-strategies using a linear program. For scalability of the framework where multiple network models of best responses are stored in the memory, we prune the weakly-dominated players' strategies to keep the oracles from becoming intractable. Finally, we conduct experiments on MNIST, CIFAR-10 and TinyImageNet for DONAS-GAN. We also evaluate the robustness under FGSM and PGD attacks on CIFAR-10, SVHN and TinyImageNet for DONAS-AT. We show that all our variants have significant improvements in both subjective qualitative evaluation and quantitative metrics, compared with their respective base architectures.

cs.LG↗

arXiv2024

Can Language Model Understand Word Semantics as A Chatbot? An Empirical Study of Language Model Internal External Mismatch

Current common interactions with language models is through full inference. This approach may not necessarily align with the model's internal knowledge. Studies show discrepancies between prompts and internal representations. Most focus on sentence understanding. We study the discrepancy of word semantics understanding in internal and external mismatch across Encoder-only, Decoder-only, and Encoder-Decoder pre-trained language models.

cs.CL↗

arXiv2024

Generative artificial intelligence for computational chemistry: a roadmap to predicting emergent phenomena

The recent surge in Generative Artificial Intelligence (AI) has introduced exciting possibilities for computational chemistry. Generative AI methods have made significant progress in sampling molecular structures across chemical species, developing force fields, and speeding up simulations. This Perspective offers a structured overview, beginning with the fundamental theoretical concepts in both Generative AI and computational chemistry. It then covers widely used Generative AI methods, including autoencoders, generative adversarial networks, reinforcement learning, flow models and language models, and highlights their selected applications in diverse areas including force field development, and protein/RNA structure prediction. A key focus is on the challenges these methods face before they become truly predictive, particularly in predicting emergent chemical phenomena. We believe that the ultimate goal of a simulation method or theory is to predict phenomena not seen before, and that Generative AI should be subject to these same standards before it is deemed useful for chemistry. We suggest that to overcome these challenges, future AI models need to integrate core chemical principles, especially from statistical mechanics.

cond-mat.stat-mech↗

arXiv2024

How Physics and Background Attributes Impact Video Transformers in Robotic Manipulation: A Case Study on Planar Pushing

As model and dataset sizes continue to scale in robot learning, the need to understand how the composition and properties of a dataset affect model performance becomes increasingly urgent to ensure cost-effective data collection and model performance. In this work, we empirically investigate how physics attributes (color, friction coefficient, shape) and scene background characteristics, such as the complexity and dynamics of interactions with background objects, influence the performance of Video Transformers in predicting planar pushing trajectories. We investigate three primary questions: How do physics attributes and background scene characteristics influence model performance? What kind of changes in attributes are most detrimental to model generalization? What proportion of fine-tuning data is required to adapt models to novel scenarios? To facilitate this research, we present CloudGripper-Push-1K, a large real-world vision-based robot pushing dataset comprising 1278 hours and 460,000 videos of planar pushing interactions with objects with different physics and background attributes. We also propose Video Occlusion Transformer (VOT), a generic modular video-transformer-based trajectory prediction framework which features 3 choices of 2D-spatial encoders as the subject of our case study. The dataset and source code are available at https://cloudgripper.org.

cs.RO↗

arXiv2024

Atomic scale insights into NaCl nucleation in nanoconfined environments

In this work we examine the nucleation from NaCl aqueous solutions within nano-confined environments, employing enhanced sampling molecular dynamics simulations integrated with machine learning-derived reaction coordinates. Through our simulations, we successfully induce phase transitions between solid, liquid, and a hydrated phase, typically observed at lower temperatures in bulk environments. Interestingly, nano-confinement serves to stabilize the solid phase and elevate melting points. Our simulations explain these findings by underscoring the significant role of water, alongside ion aggregation and subtle, anistropic dielectric behavior, in driving nucleation within nano-constrained environments. This letter thus provides a framework for sampling, analyzing and understanding nucleation processes under nano-confinement.

physics.chem-ph↗

arXiv2024

UniPredict: Large Language Models are Universal Tabular Classifiers

Tabular data prediction is a fundamental machine learning task for many applications. Existing methods predominantly employ discriminative modeling and operate under the assumption of a fixed target column, necessitating re-training for every new predictive task. Inspired by the generative power of large language models (LLMs), this paper exploits the idea of building universal tabular data predictors based on generative modeling, namely UniPredict. Here, we demonstrate the scalability of an LLM to extensive tabular datasets, enabling it to comprehend diverse tabular inputs and predict target variables following the provided instructions. Specifically, we train a single LLM on an aggregation of 169 tabular datasets with diverse targets and compare its performance against baselines that are trained on each dataset separately. We observe this versatile UniPredict model demonstrates an advantage over other models, ranging from 5.4% to 13.4%, when compared with the best tree-boosting baseline and the best neural network baseline, respectively. We further test UniPredict in few-shot learning settings on another 62 tabular datasets. Our method achieves strong performance in quickly adapting to new tasks. In low-resource few-shot setup, we observed a 100%+ performance advantage compared with XGBoost, and significant margin over all baselines. We envision that UniPredict sheds light on developing a universal tabular data prediction system that learns from data at scale and serves a wide range of prediction tasks.

cs.LG↗

arXiv2023

Is the Local Ion Density Sufficient to Drive NaCl Nucleation from the Melt and Aqueous Solution?

Even though nucleation is ubiquitous in different science and engineering problems, investigating nucleation is extremely difficult due to the complicated ranges of time and length scales involved. In this work, we simulate NaCl nucleation in both molten and aqueous environments using enhanced sampling all-atom molecular dynamics with deep learning-based estimation of reaction coordinates. By incorporating various structural order parameters and learning the reaction coordinate as a function thereof, we achieve significantly improved sampling relative to traditional ad hoc descriptions of what drives nucleation, particularly in the aqueous medium. Our results reveal a one-step nucleation mechanism in both environments, with reaction coordinate analysis highlighting the importance of local ion density in distinguishing solid and liquid states. However, while fluctuations in the local ion density are necessary to drive nucleation, they are not sufficient. Our analysis shows that near the transition states, descriptors such as enthalpy and local structure become crucial. Our protocol proposed here enables robust nucleation analysis and phase sampling, and could offer insights into nucleation mechanisms for generic small molecules in different environments.

physics.chem-ph↗

arXiv2023

Large Language Models on Lexical Semantic Change Detection: An Evaluation

Lexical Semantic Change Detection stands out as one of the few areas where Large Language Models (LLMs) have not been extensively involved. Traditional methods like PPMI, and SGNS remain prevalent in research, alongside newer BERT-based approaches. Despite the comprehensive coverage of various natural language processing domains by LLMs, there is a notable scarcity of literature concerning their application in this specific realm. In this work, we seek to bridge this gap by introducing LLMs into the domain of Lexical Semantic Change Detection. Our work presents novel prompting solutions and a comprehensive evaluation that spans all three generations of language models, contributing to the exploration of LLMs in this research area.

cs.CL↗

arXiv2023

Tree-of-Mixed-Thought: Combining Fast and Slow Thinking for Multi-hop Visual Reasoning

There emerges a promising trend of using large language models (LLMs) to generate code-like plans for complex inference tasks such as visual reasoning. This paradigm, known as LLM-based planning, provides flexibility in problem solving and endows better interpretability. However, current research is mostly limited to basic scenarios of simple questions that can be straightforward answered in a few inference steps. Planning for the more challenging multi-hop visual reasoning tasks remains under-explored. Specifically, under multi-hop reasoning situations, the trade-off between accuracy and the complexity of plan-searching becomes prominent. The prevailing algorithms either address the efficiency issue by employing the fast one-stop generation or adopt a complex iterative generation method to improve accuracy. Both fail to balance the need for efficiency and performance. Drawing inspiration from the dual system of cognition in the human brain, the fast and the slow think processes, we propose a hierarchical plan-searching algorithm that integrates the one-stop reasoning (fast) and the Tree-of-thought (slow). Our approach succeeds in performance while significantly saving inference steps. Moreover, we repurpose the PTR and the CLEVER datasets, developing a systematic framework for evaluating the performance and efficiency of LLMs-based plan-search algorithms under reasoning tasks at different levels of difficulty. Extensive experiments demonstrate the superiority of our proposed algorithm in terms of performance and efficiency. The dataset and code will be release soon.

cs.CL↗

arXiv2022

A Game-Theoretic Perspective of Generalization in Reinforcement Learning

Generalization in reinforcement learning (RL) is of importance for real deployment of RL algorithms. Various schemes are proposed to address the generalization issues, including transfer learning, multi-task learning and meta learning, as well as the robust and adversarial reinforcement learning. However, there is not a unified formulation of the various schemes, as well as the comprehensive comparisons of methods across different schemes. In this work, we propose a game-theoretic framework for the generalization in reinforcement learning, named GiRL, where an RL agent is trained against an adversary over a set of tasks, where the adversary can manipulate the distributions over tasks within a given threshold. With different configurations, GiRL can reduce the various schemes mentioned above. To solve GiRL, we adapt the widely-used method in game theory, policy space response oracle (PSRO) with the following three important modifications: i) we use model-agnostic meta learning (MAML) as the best-response oracle, ii) we propose a modified projected replicated dynamics, i.e., R-PRD, which ensures the computed meta-strategy of the adversary fall in the threshold, and iii) we also propose a protocol for the few-shot learning of the multiple strategies during testing. Extensive experiments on MuJoCo environments demonstrate that our proposed methods can outperform existing baselines, e.g., MAML.

cs.LG↗