arXiv subjects

Huiying Li

Publications and source records attributed to Huiying Li.

At least 19 recordsLinked to original sources

arXiv2026

Molt: A Scalable PyTorch-Native Training Framework for Agentic Reinforcement Learning

Agentic reinforcement learning requires rapid experimentation with agents and learning algorithms, yet large policies and long, multimodal trajectories demand substantial distributed infrastructure. We present MOLT, a lightweight, PyTorch- and Hugging Face-native framework that brings these goals together through four contributions. MOLT combines direct loading of Hugging Face models with experimentally validated trillion-parameter scalability in approximately 9.2K lines of framework code. Unified OpenAI- and Anthropic-compatible interfaces integrate existing agents with automatic handling of context compaction. Fully asynchronous training overlaps agent rollouts and policy optimization, accommodating variable agent execution times. Distributed experience storage removes centralized rollout-memory bottlenecks for long, multimodal trajectories. We experimentally validate the complete RL training pipeline on a one-trillion-parameter policy and demonstrate sustained learning with a 30B mixture-of-experts agent, establishing MOLT as a lightweight foundation for large-scale agentic RL research.

cs.LG↗

arXiv2026

Self-Supervised Skill Optimization

Agent skills provide frozen large language model (LLM) agents with reusable procedural guidance, and recent work shows that such skills can be optimized with ground-truth (GT) feedback. Many applications, however, lack GT labels, task scores, rewards, or reliable task-specific evaluators. We therefore introduce Self-Supervised Skill Optimization (SSO), a comparative framework that learns a reusable skill from unlabeled task instances alone. At each step, SSO runs the current skill on an unlabeled batch, uses a subset of the resulting executions to generate complete skill probes, and runs the probes on the same batch. An LLM judge compares the resulting answers, trajectories, artifacts, or terminal states. A separate behavior extractor identifies behavioral differences without seeing the judge's decisions. SSO uses these decisions to aggregate evidence for and against the observed behaviors across instances. It then ranks the behaviors by the resulting evidence and renders a new complete skill from the highest-ranked behaviors. The update is accepted only if the new skill outperforms the current one on an unlabeled validation set. SSO outperforms existing GT-free prompt optimizers on both closed-ended and open-ended tasks. On closed-ended benchmarks, it approaches and sometimes exceeds the strongest GT-based skill optimizer without using any GT feedback.

cs.CL↗

arXiv2026

Nemotron 3 Nano Omni: Efficient and Open Multimodal Intelligence

We introduce Nemotron 3 Nano Omni, the latest model in the Nemotron multimodal series and the first to natively support audio inputs alongside text, images, and video. Nemotron 3 Nano Omni delivers consistent accuracy improvements over its predecessor, Nemotron Nano V2 VL, across all modalities, enabled by advances in architecture, training data and recipes. In particular, Nemotron 3 delivers leading results in real-world document understanding, long audio-video comprehension, and agentic computer use. Built on the highly efficient Nemotron 3 Nano 30B-A3B backbone, Nemotron 3 Nano Omni further incorporates innovative multimodal token-reduction techniques to deliver substantially lower inference latency and higher throughput than other models of similar size. We are releasing model checkpoints in BF16, FP8, and FP4 formats, along with portions of the training data and codebase to facilitate further research and development.

cs.LG↗

arXiv2026

A hybrid statistical sampling and iterative regularization method in sparse-view computed tomography

Sparse-view computed tomography (CT) is an effective method to reduce the radiation exposure in medical imaging. To reduce the severe streaking artifacts that occur in reconstructed images due to violation of the Nyquist/Shannon sampling criterion, regularization is widely used to minimize the cost function. However, the iterative methods may lead to the accumulation and propagation of errors, which adversely affects the restoration of image details and textures. In this paper, we propose a hybrid model that integrates statistical sampling with iterative regularization to simultaneously shorten the sampling time and enhance the reconstruction quality. The proposed method is validated using three datasets: the Shepp-Logan phantom, the actual walnut X-ray projections provided by the Finnish Inverse Problems Society, and the clinical lung CT images.

physics.med-ph↗

arXiv2026

SimpleMatch: A Simple and Strong Baseline for Semantic Correspondence

Recent advances in semantic correspondence have been largely driven by the use of pre-trained large-scale models. However, a limitation of these approaches is their dependence on high-resolution input images to achieve optimal performance, which results in considerable computational overhead. In this work, we address a fundamental limitation in current methods: the irreversible fusion of adjacent keypoint features caused by deep downsampling operations. This issue is triggered when semantically distinct keypoints fall within the same downsampled receptive field (e.g., 16x16 patches). To address this issue, we present SimpleMatch, a simple yet effective framework for semantic correspondence that delivers strong performance even at low resolutions. We propose a lightweight upsample decoder that progressively recovers spatial detail by upsampling deep features to 1/4 resolution, and a multi-scale supervised loss that ensures the upsampled features retain discriminative features across different spatial scales. In addition, we introduce sparse matching and window-based localization to optimize training memory usage and reduce it by 51%. At a resolution of 252x252 (3.3x smaller than current SOTA methods), SimpleMatch achieves superior performance with 84.1% PCK@0.1 on the SPair-71k benchmark. We believe this framework provides a practical and efficient baseline for future research in semantic correspondence. Code is available at: https://github.com/hailong23-jin/SimpleMatch.

cs.CV↗

arXiv2025

TongSIM: A General Platform for Simulating Intelligent Machines

As artificial intelligence (AI) rapidly advances, especially in multimodal large language models (MLLMs), research focus is shifting from single-modality text processing to the more complex domains of multimodal and embodied AI. Embodied intelligence focuses on training agents within realistic simulated environments, leveraging physical interaction and action feedback rather than conventionally labeled datasets. Yet, most existing simulation platforms remain narrowly designed, each tailored to specific tasks. A versatile, general-purpose training environment that can support everything from low-level embodied navigation to high-level composite activities, such as multi-agent social simulation and human-AI collaboration, remains largely unavailable. To bridge this gap, we introduce TongSIM, a high-fidelity, general-purpose platform for training and evaluating embodied agents. TongSIM offers practical advantages by providing over 100 diverse, multi-room indoor scenarios as well as an open-ended, interaction-rich outdoor town simulation, ensuring broad applicability across research needs. Its comprehensive evaluation framework and benchmarks enable precise assessment of agent capabilities, such as perception, cognition, decision-making, human-robot cooperation, and spatial and social reasoning. With features like customized scenes, task-adaptive fidelity, diverse agent types, and dynamic environmental simulation, TongSIM delivers flexibility and scalability for researchers, serving as a unified platform that accelerates training, evaluation, and advancement toward general embodied intelligence.

cs.AI↗

arXiv2025

Question Answering Over Spatio-Temporal Knowledge Graph

Spatio-temporal knowledge graphs (STKGs) enhance traditional KGs by integrating temporal and spatial annotations, enabling precise reasoning over questions with spatio-temporal dependencies. Despite their potential, research on spatio-temporal knowledge graph question answering (STKGQA) remains limited. This is primarily due to the lack of datasets that simultaneously contain spatio-temporal information, as well as methods capable of handling implicit spatio-temporal reasoning. To bridge this gap, we introduce the spatio-temporal question answering dataset (STQAD), the first comprehensive benchmark comprising 10,000 natural language questions that require both temporal and spatial reasoning. STQAD is constructed with real-world facts containing spatio-temporal information, ensuring that the dataset reflects practical scenarios. Furthermore, our experiments reveal that existing KGQA methods underperform on STQAD, primarily due to their inability to model spatio-temporal interactions. To address this, we propose the spatio-temporal complex question answering (STCQA) method, which jointly embeds temporal and spatial features into KG representations and dynamically filters answers through constraint-aware reasoning. STCQA achieves state-of-the-art performance, significantly outperforming existing baselines. Our work not only provides a valuable resource for future research but also advances the field by offering a robust baseline for answering complex spatio-temporal questions.

cs.CL↗

arXiv2025

CollabVLA: Self-Reflective Vision-Language-Action Model Dreaming Together with Human

In this work, we present CollabVLA, a self-reflective vision-language-action framework that transforms a standard visuomotor policy into a collaborative assistant. CollabVLA tackles key limitations of prior VLAs, including domain overfitting, non-interpretable reasoning, and the high latency of auxiliary generative models, by integrating VLM-based reflective reasoning with diffusion-based action generation under a mixture-of-experts design. Through a two-stage training recipe of action grounding and reflection tuning, it supports explicit self-reflection and proactively solicits human guidance when confronted with uncertainty or repeated failure. It cuts normalized Time by ~2x and Dream counts by ~4x vs. generative agents, achieving higher success rates, improved interpretability, and balanced low latency compared with existing methods. This work takes a pioneering step toward shifting VLAs from opaque controllers to genuinely assistive agents capable of reasoning, acting, and collaborating with humans.

cs.RO↗

arXiv2025

PRE-MAP: Personalized Reinforced Eye-tracking Multimodal LLM for High-Resolution Multi-Attribute Point Prediction

Visual selective attention, driven by individual preferences, regulates human prioritization of visual stimuli by bridging subjective cognitive mechanisms with objective visual elements, thereby steering the semantic interpretation and hierarchical processing of dynamic visual scenes. However, existing models and datasets predominantly neglect the influence of subjective cognitive diversity on fixation behavior. Conventional saliency prediction models, typically employing segmentation approaches, rely on low-resolution imagery to generate saliency heatmaps, subsequently upscaled to native resolutions, which limiting their capacity to capture personalized attention patterns. Furthermore, MLLMs are constrained by factors such as hallucinations, making it very costly to strictly adhere to the expected format in tasks involving multiple point predictions, and achieving precise point positioning is challenging. To address these limitations, we present Subjective Personalized Attention for Advertisement Videos, namely SPA-ADV, a large-scale multimodal dataset capturing gaze behaviors from over 4,500 participants varying in age and gender with 486 videos. Furthermore, we propose PRE-MAP, a novel eye-tracking saliency model that characterizes Personalized visual disparities through Reinforcement learning-optimized Eye-tracking, built upon MLLMs and guided by Multi-Attribute user profiles to predict Points. To ensure MLLMs produce prediction points that are both format-correct and spatially accurate, we introduce Consistency Group Relative Policy Optimization (C-GRPO), inspired by the variability in eye movement points and Multi-Attribute profiles. Extensive experiments on SPA-ADV and other benchmarks demonstrate the effectiveness of our approach. The code and dataset are available at \href{https://github.com/mininglamp-MLLM/PRE-MAP}{this URL}.

cs.CV↗

arXiv2025

Body Discovery of Embodied AI

In the pursuit of realizing artificial general intelligence (AGI), the importance of embodied artificial intelligence (AI) becomes increasingly apparent. Following this trend, research integrating robots with AGI has become prominent. As various kinds of embodiments have been designed, adaptability to diverse embodiments will become important to AGI. We introduce a new challenge, termed "Body Discovery of Embodied AI", focusing on tasks of recognizing embodiments and summarizing neural signal functionality. The challenge encompasses the precise definition of an AI body and the intricate task of identifying embodiments in dynamic environments, where conventional approaches often prove inadequate. To address these challenges, we apply causal inference method and evaluate it by developing a simulator tailored for testing algorithms with virtual environments. Finally, we validate the efficacy of our algorithms through empirical testing, demonstrating their robust performance in various scenarios based on virtual environments.

cs.RO↗

arXiv2024

Bounding-Box Inference for Error-Aware Model-Based Reinforcement Learning

In model-based reinforcement learning, simulated experiences from the learned model are often treated as equivalent to experience from the real environment. However, when the model is inaccurate, it can catastrophically interfere with policy learning. Alternatively, the agent might learn about the model's accuracy and selectively use it only when it can provide reliable predictions. We empirically explore model uncertainty measures for selective planning and show that best results require distribution insensitive inference to estimate the uncertainty over model-based updates. To that end, we propose and evaluate bounding-box inference, which operates on bounding-boxes around sets of possible states and other quantities. We find that bounding-box inference can reliably support effective selective planning.

cs.LG↗

arXiv2024

Independently Keypoint Learning for Small Object Semantic Correspondence

Semantic correspondence remains a challenging task for establishing correspondences between a pair of images with the same category or similar scenes due to the large intra-class appearance. In this paper, we introduce a novel problem called 'Small Object Semantic Correspondence (SOSC).' This problem is challenging due to the close proximity of keypoints associated with small objects, which results in the fusion of these respective features. It is difficult to identify the corresponding key points of the fused features, and it is also difficult to be recognized. To address this challenge, we propose the Keypoint Bounding box-centered Cropping (KBC) method, which aims to increase the spatial separation between keypoints of small objects, thereby facilitating independent learning of these keypoints. The KBC method is seamlessly integrated into our proposed inference pipeline and can be easily incorporated into other methodologies, resulting in significant performance enhancements. Additionally, we introduce a novel framework, named KBCNet, which serves as our baseline model. KBCNet comprises a Cross-Scale Feature Alignment (CSFA) module and an efficient 4D convolutional decoder. The CSFA module is designed to align multi-scale features, enriching keypoint representations by integrating fine-grained features and deep semantic features. Meanwhile, the 4D convolutional decoder, based on efficient 4D convolution, ensures efficiency and rapid convergence. To empirically validate the effectiveness of our proposed methodology, extensive experiments are conducted on three widely used benchmarks: PF-PASCAL, PF-WILLOW, and SPair-71k. Our KBC method demonstrates a substantial performance improvement of 7.5\% on the SPair-71K dataset, providing compelling evidence of its efficacy.

cs.CV↗

arXiv2023

On the Permanence of Backdoors in Evolving Models

Existing research on training-time attacks for deep neural networks (DNNs), such as backdoors, largely assume that models are static once trained, and hidden backdoors trained into models remain active indefinitely. In practice, models are rarely static but evolve continuously to address distribution drifts in the underlying data. This paper explores the behavior of backdoor attacks in time-varying models, whose model weights are continually updated via fine-tuning to adapt to data drifts. Our theoretical analysis shows how fine-tuning with fresh data progressively "erases" the injected backdoors, and our empirical study illustrates how quickly a time-varying model "forgets" backdoors under a variety of training and attack settings. We also show that novel fine-tuning strategies using smart learning rates can significantly accelerate backdoor forgetting. Finally, we discuss the need for new backdoor defenses that target time-varying models specifically.

cs.CR↗

arXiv2022

Outlining and Filling: Hierarchical Query Graph Generation for Answering Complex Questions over Knowledge Graphs

Query graph construction aims to construct the correct executable SPARQL on the KG to answer natural language questions. Although recent methods have achieved good results using neural network-based query graph ranking, they suffer from three new challenges when handling more complex questions: 1) complicated SPARQL syntax, 2) huge search space, and 3) locally ambiguous query graphs. In this paper, we provide a new solution. As a preparation, we extend the query graph by treating each SPARQL clause as a subgraph consisting of vertices and edges and define a unified graph grammar called AQG to describe the structure of query graphs. Based on these concepts, we propose a novel end-to-end model that performs hierarchical autoregressive decoding to generate query graphs. The high-level decoding generates an AQG as a constraint to prune the search space and reduce the locally ambiguous query graph. The bottom-level decoding accomplishes the query graph construction by selecting appropriate instances from the preprepared candidates to fill the slots in the AQG. The experimental results show that our method greatly improves the SOTA performance on complex KGQA benchmarks. Equipped with pre-trained models, the performance of our method is further improved, achieving SOTA for all three datasets used.

cs.AI↗

arXiv2022

Blacklight: Scalable Defense for Neural Networks against Query-Based Black-Box Attacks

Deep learning systems are known to be vulnerable to adversarial examples. In particular, query-based black-box attacks do not require knowledge of the deep learning model, but can compute adversarial examples over the network by submitting queries and inspecting returns. Recent work largely improves the efficiency of those attacks, demonstrating their practicality on today's ML-as-a-service platforms. We propose Blacklight, a new defense against query-based black-box adversarial attacks. The fundamental insight driving our design is that, to compute adversarial examples, these attacks perform iterative optimization over the network, producing image queries highly similar in the input space. Blacklight detects query-based black-box attacks by detecting highly similar queries, using an efficient similarity engine operating on probabilistic content fingerprints. We evaluate Blacklight against eight state-of-the-art attacks, across a variety of models and image classification tasks. Blacklight identifies them all, often after only a handful of queries. By rejecting all detected queries, Blacklight prevents any attack to complete, even when attackers persist to submit queries after account ban or query rejection. Blacklight is also robust against several powerful countermeasures, including an optimal black-box attack that approximates white-box attacks in efficiency. Finally, we illustrate how Blacklight generalizes to other domains like text classification.

cs.CR↗

arXiv2021

Leveraging Table Content for Zero-shot Text-to-SQL with Meta-Learning

Single-table text-to-SQL aims to transform a natural language question into a SQL query according to one single table. Recent work has made promising progress on this task by pre-trained language models and a multi-submodule framework. However, zero-shot table, that is, the invisible table in the training set, is currently the most critical bottleneck restricting the application of existing approaches to real-world scenarios. Although some work has utilized auxiliary tasks to help handle zero-shot tables, expensive extra manual annotation limits their practicality. In this paper, we propose a new approach for the zero-shot text-to-SQL task which does not rely on any additional manual annotations. Our approach consists of two parts. First, we propose a new model that leverages the abundant information of table content to help establish the mapping between questions and zero-shot tables. Further, we propose a simple but efficient meta-learning strategy to train our model. The strategy utilizes the two-step gradient update to force the model to learn a generalization ability towards zero-shot tables. We conduct extensive experiments on a public open-domain text-to-SQL dataset WikiSQL and a domain-specific dataset ESQL. Compared to existing approaches using the same pre-trained model, our approach achieves significant improvements on both datasets. Compared to the larger pre-trained model and the tabular-specific pre-trained model, our approach is still competitive. More importantly, on the zero-shot subsets of both the datasets, our approach further increases the improvements.

cs.CL↗

arXiv2021

Formal Query Building with Query Structure Prediction for Complex Question Answering over Knowledge Base

Formal query building is an important part of complex question answering over knowledge bases. It aims to build correct executable queries for questions. Recent methods try to rank candidate queries generated by a state-transition strategy. However, this candidate generation strategy ignores the structure of queries, resulting in a considerable number of noisy queries. In this paper, we propose a new formal query building approach that consists of two stages. In the first stage, we predict the query structure of the question and leverage the structure to constrain the generation of the candidate queries. We propose a novel graph generation framework to handle the structure prediction task and design an encoder-decoder model to predict the argument of the predetermined operation in each generative step. In the second stage, we follow the previous methods to rank the candidate queries. The experimental results show that our formal query building approach outperforms existing methods on complex questions while staying competitive on simple questions.

cs.CL↗

arXiv2020

Piracy Resistant Watermarks for Deep Neural Networks

As companies continue to invest heavily in larger, more accurate and more robust deep learning models, they are exploring approaches to monetize their models while protecting their intellectual property. Model licensing is promising, but requires a robust tool for owners to claim ownership of models, i.e. a watermark. Unfortunately, current designs have not been able to address piracy attacks, where third parties falsely claim model ownership by embedding their own "pirate watermarks" into an already-watermarked model. We observe that resistance to piracy attacks is fundamentally at odds with the current use of incremental training to embed watermarks into models. In this work, we propose null embedding, a new way to build piracy-resistant watermarks into DNNs that can only take place at a model's initial training. A null embedding takes a bit string (watermark value) as input, and builds strong dependencies between the model's normal classification accuracy and the watermark. As a result, attackers cannot remove an embedded watermark via tuning or incremental training, and cannot add new pirate watermarks to already watermarked models. We empirically show that our proposed watermarks achieve piracy resistance and other watermark properties, over a wide range of tasks and models. Finally, we explore a number of adaptive counter-measures, and show our watermark remains robust against a variety of model modifications, including model fine-tuning, compression, and existing methods to detect/remove backdoors. Our watermarked models are also amenable to transfer learning without losing their watermark properties.

cs.CR↗