arXiv subjects

Xiangrui Zeng

Publications and source records attributed to Xiangrui Zeng.

At least 19 recordsLinked to original sources

arXiv2026

SimpleMemVLA: A Simple but Effective Native-Video Memory for Vision-Language-Action Models

Long-horizon manipulation is partially observable: the information needed to choose the next action may appear only in observations from minutes earlier. Existing memory mechanisms: retrieval banks, learned compressors, recurrent states must decide what to keep from the past before knowing what a future decision will require. This was motivated by the assumption that minute-scale history is too large to process directly, which modern VLM backbones no longer make true. In this work, we introduce SimpleMemVLA, a VLA without a dedicated memory module. It keeps the sampled history intact and passes it to the backbone in the timestamped video format the backbone was pretrained to process; the hidden states of a generated sub-task then form the only channel from history to a standard flow-matching action head. Since consecutive decisions share most of their history, prefilling the shared prefix during action execution keeps latency close to a single-frame VLA. SimpleMemVLA sets a new state of the art on four memory benchmarks without cost on general-purpose control. Holding the backbone and training setup fixed, it outperforms retrieval, compression and recurrent-state mechanisms by a wide margin, and causal interventions confirm that the policy genuinely reads its history. Code available at https://github.com/wadeKeith/SimpleMemVLA

cs.CV↗

arXiv2026

Chronos: A Physics-Informed Full-History Framework for Non-Markovian Long-Horizon Manipulation

General-purpose robot policies should be modeled as dynamical systems, yet many VLA and generative imitation policies still rely on present observations or short windows. This Markovian shortcut fails in memory-dependent manipulation: identical observations can demand different actions after different histories. We present Chronos, a physics-informed full-history framework for non-Markovian long-horizon manipulation. The key idea is to elevate observation history from auxiliary context to the latent state of the policy dynamics. At each physical control step, Chronos forms one state-representative token by fusing observation and proprioception, so the token sequence is aligned one-to-one with physical time. A selective state space model propagates this causal historical state, which conditions a multimodal coarse action prior through implicit maximum likelihood estimation (IMLE). This prior is then refined by a second-order Schrodinger-inspired bridge that predicts acceleration fields, yielding smoother and more physically grounded robot motion. Across 16 simulated tasks and 4 real-world experiments, Chronos is evaluated on precision insertion, general manipulation, and memory-dependent long-horizon control. On RMBench, where success requires remembering task phase, Chronos achieves 73.6% average success, outperforming Markovian VLA baseline pi0.5 by +62.4 percentage points, a 6.6x relative gain, while using 10x fewer parameters. It also surpasses the memory VLA Mem-0 by 22.8 points while using over 30x fewer parameters. In real-world dual-arm experiments using a single RGB camera, Chronos achieves 78% average success over four tasks, including 72% on the three memory-dependent tasks, whereas pi0.5 achieves 7% overall and 0% on the memory-dependent subset. These results suggest that history should not be treated as auxiliary context, but as the latent state of the manipulation policy.

cs.RO↗

arXiv2026

Humanizing Robot Gaze Shifts: A Framework for Natural Gaze Shifts in Humanoid Robots

Leveraging auditory and visual feedback for attention reorientation is essential for natural gaze shifts in social interaction. However, enabling humanoid robots to perform natural and context-appropriate gaze shifts in unconstrained human--robot interaction (HRI) remains challenging, as it requires the coupling of cognitive attention mechanisms and biomimetic motion generation. In this work, we propose the Robot Gaze-Shift (RGS) framework, which integrates these two components into a unified pipeline. First, RGS employs a vision--language model (VLM)-based gaze reasoning pipeline to infer context-appropriate gaze targets from multimodal interaction cues, ensuring consistency with human gaze-orienting regularities. Second, RGS introduces a conditional Vector Quantized-Variational Autoencoder (VQ-VAE) model for eye--head coordinated gaze-shift motion generation, producing diverse and human-like gaze-shift behaviors. Experiments validate that RGS effectively replicates human-like target selection and generates realistic, diverse gaze-shift motions.

cs.RO↗

arXiv2025

DeepThinkVLA: Enhancing Reasoning Capability of Vision-Language-Action Models

Does Chain-of-Thought (CoT) reasoning genuinely improve Vision Language Action (VLA) models, or does it merely add overhead? Existing CoT-VLA systems report limited and inconsistent gains, yet no prior work has rigorously diagnosed when and why CoT helps robots act. Through systematic experiments, we identify two necessary conditions that must be jointly satisfied for CoT to be effective in VLA: (1) Decoding Alignment: CoT and actions must be generated with modality-appropriate mechanisms; forcing both through a single autoregressive decoder is not merely suboptimal but actively harmful, degrading performance by 4.2 percentage points; (2) Causal Alignment: CoT must be causally linked to task success via outcome-based optimization; without it, supervised CoT is indistinguishable from no reasoning at all under action-execution-sensitive dynamics shift, exhibiting a 32.0 pp performance drop nearly identical to the 31.6 pp drop of a reasoning-free baseline. Guided by these findings, we build DeepThinkVLA: a hybrid-attention decoder satisfies Condition 1 by pairing causal attention for language with bidirectional attention for parallel action decoding, while a two-stage SFT-then-RL pipeline satisfies Condition 2 by aligning the full reasoning: action chain with sparse task-success rewards. DeepThinkVLA achieves 97.0\% success on LIBERO, 79.0\% robustness on LIBERO-Plus (vs. 61.6\% for $\pi_0$-FAST), and 59.3\% success on RoboTwin 2.0, exceeding the strongest baseline by 21.7 points. Furthermore, real-robot experiments provide preliminary evidence for the physical applicability of our CoT data construction and hybrid architecture. Our codes are available at https://github.com/OpenBMB/DeepThinkVLA.

cs.LG↗

arXiv2025

Leveraging Diffusion Knowledge for Generative Image Compression with Fractal Frequency-Aware Band Learning

By optimizing the rate-distortion-realism trade-off, generative image compression approaches produce detailed, realistic images instead of the only sharp-looking reconstructions produced by rate-distortion-optimized models. In this paper, we propose a novel deep learning-based generative image compression method injected with diffusion knowledge, obtaining the capacity to recover more realistic textures in practical scenarios. Efforts are made from three perspectives to navigate the rate-distortion-realism trade-off in the generative image compression task. First, recognizing the strong connection between image texture and frequency-domain characteristics, we design a Fractal Frequency-Aware Band Image Compression (FFAB-IC) network to effectively capture the directional frequency components inherent in natural images. This network integrates commonly used fractal band feature operations within a neural non-linear mapping design, enhancing its ability to retain essential given information and filter out unnecessary details. Then, to improve the visual quality of image reconstruction under limited bandwidth, we integrate diffusion knowledge into the encoder and implement diffusion iterations into the decoder process, thus effectively recovering lost texture details. Finally, to fully leverage the spatial and frequency intensity information, we incorporate frequency- and content-aware regularization terms to regularize the training of the generative image compression network. Extensive experiments in quantitative and qualitative evaluations demonstrate the superiority of the proposed method, advancing the boundaries of achievable distortion-realism pairs, i.e., our method achieves better distortions at high realism and better realism at low distortion than ever before.

cs.CV↗

arXiv2024

Learn2Synth: Learning Optimal Data Synthesis Using Hypergradients for Brain Image Segmentation

Domain randomization through synthesis is a powerful strategy to train networks that are unbiased with respect to the domain of the input images. Randomization allows networks to see a virtually infinite range of intensities and artifacts during training, thereby minimizing overfitting to appearance and maximizing generalization to unseen data. Although powerful, this approach relies on the accurate tuning of a large set of hyperparameters that govern the probabilistic distribution of the synthesized images. Instead of manually tuning these parameters, we introduce Learn2Synth, a novel procedure in which synthesis parameters are learned using a small set of real labeled data. Unlike methods that impose constraints to align synthetic data with real data (e.g., contrastive or adversarial techniques), which risk misaligning the image and its label map, we tune an augmentation engine such that a segmentation network trained on synthetic data has optimal accuracy when applied to real data. This approach allows the training procedure to benefit from real labeled examples, without ever using these real examples to train the segmentation network, which avoids biasing the network towards the properties of the training set. Specifically, we develop parametric and nonparametric strategies to enhance synthetic images in a way that improves the performance of the segmentation network. We demonstrate the effectiveness of this learning strategy on synthetic and real-world brain scans. Code is available at: https://github.com/HuXiaoling/Learn2Synth.

cs.CV↗

arXiv2024

Does Optimal Control Always Benefit from Better Prediction? An Analysis Framework for Predictive Optimal Control

The ``prediction + optimal control'' scheme has shown good performance in many applications of automotive, traffic, robot, and building control. In practice, the prediction results are simply considered correct in the optimal control design process. However, in reality, these predictions may never be perfect. Under a conventional stochastic optimal control formulation, it is difficult to answer questions like ``what if the predictions are wrong''. This paper presents an analysis framework for predictive optimal control where the subjective belief about the future is no longer considered perfect. A novel concept called the hidden prediction state is proposed to establish connections among the predictors, the subjective beliefs, the control policies and the objective control performance. Based on this framework, the predictor evaluation problem is analyzed. Three commonly-used predictor evaluation measures, including the mean squared error, the regret and the log-likelihood, are considered. It is shown that neither using the mean square error nor using the likelihood can guarantee a monotonic relationship between the predictor error and the optimal control cost. To guarantee control cost improvement, it is suggested the predictor should be evaluated with the control performance, e.g., using the optimal control cost or the regret to evaluate predictors. Numerical examples and examples from automotive applications with real-world driving data are provided to illustrate the ideas and the results.

eess.SY↗

arXiv2022

SHREC 2021: Classification in cryo-electron tomograms

Cryo-electron tomography (cryo-ET) is an imaging technique that allows three-dimensional visualization of macro-molecular assemblies under near-native conditions. Cryo-ET comes with a number of challenges, mainly low signal-to-noise and inability to obtain images from all angles. Computational methods are key to analyze cryo-electron tomograms. To promote innovation in computational methods, we generate a novel simulated dataset to benchmark different methods of localization and classification of biological macromolecules in tomograms. Our publicly available dataset contains ten tomographic reconstructions of simulated cell-like volumes. Each volume contains twelve different types of complexes, varying in size, function and structure. In this paper, we have evaluated seven different methods of finding and classifying proteins. Seven research groups present results obtained with learning-based methods and trained on the simulated dataset, as well as a baseline template matching (TM), a traditional method widely used in cryo-ET research. We show that learning-based approaches can achieve notably better localization and classification performance than TM. We also experimentally confirm that there is a negative relationship between particle size and performance for all methods.

eess.IV↗

arXiv2021

Boosting Active Learning via Improving Test Performance

Central to active learning (AL) is what data should be selected for annotation. Existing works attempt to select highly uncertain or informative data for annotation. Nevertheless, it remains unclear how selected data impacts the test performance of the task model used in AL. In this work, we explore such an impact by theoretically proving that selecting unlabeled data of higher gradient norm leads to a lower upper-bound of test loss, resulting in better test performance. However, due to the lack of label information, directly computing gradient norm for unlabeled data is infeasible. To address this challenge, we propose two schemes, namely expected-gradnorm and entropy-gradnorm. The former computes the gradient norm by constructing an expected empirical loss while the latter constructs an unsupervised loss with entropy. Furthermore, we integrate the two schemes in a universal AL framework. We evaluate our method on classical image classification and semantic segmentation tasks. To demonstrate its competency in domain applications and its robustness to noise, we also validate our method on a cellular imaging analysis task, namely cryo-Electron Tomography subtomogram classification. Results demonstrate that our method achieves superior performance against the state of the art. Our source code is available at https://github.com/xulabs/aitom/blob/master/doc/projects/al_gradnorm.md.

cs.LG↗

arXiv2021

Cryo-shift: Reducing domain shift in cryo-electron subtomograms with unsupervised domain adaptation and randomization

Cryo-Electron Tomography (cryo-ET) is a 3D imaging technology that enables the visualization of subcellular structures in situ at near-atomic resolution. Cellular cryo-ET images help in resolving the structures of macromolecules and determining their spatial relationship in a single cell, which has broad significance in cell and structural biology. Subtomogram classification and recognition constitute a primary step in the systematic recovery of these macromolecular structures. Supervised deep learning methods have been proven to be highly accurate and efficient for subtomogram classification, but suffer from limited applicability due to scarcity of annotated data. While generating simulated data for training supervised models is a potential solution, a sizeable difference in the image intensity distribution in generated data as compared to real experimental data will cause the trained models to perform poorly in predicting classes on real subtomograms. In this work, we present Cryo-Shift, a fully unsupervised domain adaptation and randomization framework for deep learning-based cross-domain subtomogram classification. We use unsupervised multi-adversarial domain adaption to reduce the domain shift between features of simulated and experimental data. We develop a network-driven domain randomization procedure with `warp' modules to alter the simulated data and help the classifier generalize better on experimental data. We do not use any labeled experimental data to train our model, whereas some of the existing alternative approaches require labeled experimental samples for cross-domain classification. Nevertheless, Cryo-Shift outperforms the existing alternative approaches in cross-domain subtomogram classification in extensive evaluation studies demonstrated herein using both simulated and experimental data.

q-bio.QM↗

arXiv2021

Disentangling semantic features of macromolecules in Cryo-Electron Tomography

Cryo-electron tomography (Cryo-ET) is a 3D imaging technique that enables the systemic study of shape, abundance, and distribution of macromolecular structures in single cells in near-atomic resolution. However, the systematic and efficient $\textit{de novo}$ recognition and recovery of macromolecular structures captured by Cryo-ET are very challenging due to the structural complexity and imaging limits. Even macromolecules with identical structures have various appearances due to different orientations and imaging limits, such as noise and the missing wedge effect. Explicitly disentangling the semantic features of macromolecules is crucial for performing several downstream analyses on the macromolecules. This paper has addressed the problem by proposing a 3D Spatial Variational Autoencoder that explicitly disentangle the structure, orientation, and shift of macromolecules. Extensive experiments on both synthesized and real cryo-ET datasets and cross-domain evaluations demonstrate the efficacy of our method.

q-bio.BM↗

arXiv2021

Predictive Optimal Control with Data-Based Disturbance Scenario Tree Approximation

Efficiently computing the optimal control policy concerning a complicated future with stochastic disturbance has always been a challenge. The predicted stochastic future disturbance can be represented by a scenario tree, but solving the optimal control problem with a scenario tree is usually computationally demanding. In this paper, we propose a data-based clustering approximation method for the scenario tree representation. Differently from the popular Markov chain approximation, the proposed method can retain information from previous steps while keeping the state space size small. Then the predictive optimal control problem can be approximately solved with reduced computational load using dynamic programming. The proposed method is evaluated in numerical examples and compared with the method which considers the disturbance as a non-stationary Markov chain. The results show that the proposed method can achieve better control performance than the Markov chain method.

eess.SY↗

arXiv2021

Active Learning to Classify Macromolecular Structures in situ for Less Supervision in Cryo-Electron Tomography

Motivation: Cryo-Electron Tomography (cryo-ET) is a 3D bioimaging tool that visualizes the structural and spatial organization of macromolecules at a near-native state in single cells, which has broad applications in life science. However, the systematic structural recognition and recovery of macromolecules captured by cryo-ET are difficult due to high structural complexity and imaging limits. Deep learning based subtomogram classification have played critical roles for such tasks. As supervised approaches, however, their performance relies on sufficient and laborious annotation on a large training dataset. Results: To alleviate this major labeling burden, we proposed a Hybrid Active Learning (HAL) framework for querying subtomograms for labelling from a large unlabeled subtomogram pool. Firstly, HAL adopts uncertainty sampling to select the subtomograms that have the most uncertain predictions. Moreover, to mitigate the sampling bias caused by such strategy, a discriminator is introduced to judge if a certain subtomogram is labeled or unlabeled and subsequently the model queries the subtomogram that have higher probabilities to be unlabeled. Additionally, HAL introduces a subset sampling strategy to improve the diversity of the query set, so that the information overlap is decreased between the queried batches and the algorithmic efficiency is improved. Our experiments on subtomogram classification tasks using both simulated and real data demonstrate that we can achieve comparable testing performance (on average only 3% accuracy drop) by using less than 30% of the labeled subtomograms, which shows a very promising result for subtomogram classification task with limited labeling resources.

q-bio.QM↗

arXiv2020

Feedback-Based Dynamic Feature Selection for Constrained Continuous Data Acquisition

Relevant and high-quality data are critical to successful development of machine learning applications. For machine learning applications on dynamic systems equipped with a large number of sensors, such as connected vehicles and robots, how to find relevant and high-quality data features in an efficient way is a challenging problem. In this work, we address the problem of feature selection in constrained continuous data acquisition. We propose a feedback-based dynamic feature selection algorithm that efficiently decides on the feature set for data collection from a dynamic system in a step-wise manner. We formulate the sequential feature selection procedure as a Markov Decision Process. The machine learning model performance feedback with an exploration component is used as the reward function in an $\epsilon$-greedy action selection. Our evaluation shows that the proposed feedback-based feature selection algorithm has superior performance over constrained baseline methods and matching performance with unconstrained baseline methods.

cs.LG↗

arXiv2020

Few shot domain adaptation for in situ macromolecule structural classification in cryo-electron tomograms

Motivation: Cryo-Electron Tomography (cryo-ET) visualizes structure and spatial organization of macromolecules and their interactions with other subcellular components inside single cells in the close-to-native state at sub-molecular resolution. Such information is critical for the accurate understanding of cellular processes. However, subtomogram classification remains one of the major challenges for the systematic recognition and recovery of the macromolecule structures in cryo-ET because of imaging limits and data quantity. Recently, deep learning has significantly improved the throughput and accuracy of large-scale subtomogram classification. However often it is difficult to get enough high-quality annotated subtomogram data for supervised training due to the enormous expense of labeling. To tackle this problem, it is beneficial to utilize another already annotated dataset to assist the training process. However, due to the discrepancy of image intensity distribution between source domain and target domain, the model trained on subtomograms in source domainmay perform poorly in predicting subtomogram classes in the target domain. Results: In this paper, we adapt a few shot domain adaptation method for deep learning based cross-domain subtomogram classification. The essential idea of our method consists of two parts: 1) take full advantage of the distribution of plentiful unlabeled target domain data, and 2) exploit the correlation between the whole source domain dataset and few labeled target domain data. Experiments conducted on simulated and real datasets show that our method achieves significant improvement on cross domain subtomogram classification compared with baseline methods.

q-bio.QM↗

arXiv2019

AITom: Open-source AI platform for cryo-electron tomography data analysis

Cryo-electron tomography (cryo-ET) is an emerging technology for the 3D visualization of structural organizations and interactions of subcellular components at near-native state and sub-molecular resolution. Tomograms captured by cryo-ET contain heterogeneous structures representing the complex and dynamic subcellular environment. Since the structures are not purified or fluorescently labeled, the spatial organization and interaction between both the known and unknown structures can be studied in their native environment. The rapid advances of cryo-electron tomography (cryo-ET) have generated abundant 3D cellular imaging data. However, the systematic localization, identification, segmentation, and structural recovery of the subcellular components require efficient and accurate large-scale image analysis methods. We introduce AITom, an open-source artificial intelligence platform for cryo-ET researchers. AITom provides many public as well as in-house algorithms for performing cryo-ET data analysis through both the traditional template-based or template-free approach and the deep learning approach. AITom also supports remote interactive analysis. Comprehensive tutorials for each analysis module are provided to guide the user through. We welcome researchers and developers to join this collaborative open-source software development project. Availability: https://github.com/xulabs/aitom

q-bio.QM↗

arXiv2019

CS Sparse K-means: An Algorithm for Cluster-Specific Feature Selection in High-Dimensional Clustering

Feature selection is an important and challenging task in high dimensional clustering. For example, in genomics, there may only be a small number of genes that are differentially expressed, which are informative to the overall clustering structure. Existing feature selection methods, such as Sparse K-means, rarely tackle the problem of accounting features that can only separate a subset of clusters. In genomics, it is highly likely that a gene can only define one subtype against all the other subtypes or distinguish a pair of subtypes but not others. In this paper, we propose a K-means based clustering algorithm that discovers informative features as well as which cluster pairs are separable by each selected features. The method is essentially an EM algorithm, in which we introduce lasso-type constraints on each cluster pair in the M step, and make the E step possible by maximizing the raw cross-cluster distance instead of minimizing the intra-cluster distance. The results were demonstrated on simulated data and a leukemia gene expression dataset.

stat.ME↗

arXiv2019

Simultaneous Estimation of Number of Clusters and Feature Sparsity in Clustering High-Dimensional Data

Estimating the number of clusters (K) is a critical and often difficult task in cluster analysis. Many methods have been proposed to estimate K, including some top performers using resampling approach. When performing cluster analysis in high-dimensional data, simultaneous clustering and feature selection is needed for improved interpretation and performance. To our knowledge, none has investigated simultaneous estimation of K and feature selection in an exploratory cluster analysis. In this paper, we propose a resampling method to meet this gap and evaluate its performance under the sparse K-means clustering framework. The proposed target function balances between sensitivity and specificity of clustering evaluation of pairwise subjects from clustering of full and subsampled data. Through extensive simulations, the method performs among the best over classical methods in estimating K in low-dimensional data. For high-dimensional simulation data, it also shows superior performance to simultaneously estimate K and feature sparsity parameter. Finally, we evaluated the methods in four microarray, two RNA-seq, one SNP and two non-omics datasets. The proposed method achieves better clustering accuracy with fewer selected predictive genes in almost all real applications.

stat.ME↗