Searcharxiv⌕ Search

arXiv subjects

Hongbin Liu

Publications and source records attributed to Hongbin Liu.

At least 19 recordsLinked to original sources

Scintillator-Based Event-Mode Neutron Imaging with Timepix4 Readout

Scintillator-based event-mode neutron imaging is emerging as a powerful approach to neutron radiography. Unlike conventional integrating-mode neutron imaging, which accumulates scintillation light during an exposure, event reconstruction estimates neutron-event positions and arrival times from time-stamped optical responses. We investigated this approach using a self-developed Timepix4-based optical readout system, image-formation analysis, and beam experiments at the China Spallation Neutron Source under different scintillator and optical configurations. The analysis explains how event localization can reduce image blur associated with the spatial extent of individual scintillation responses. Under the same 0.35$\times$ optical configuration, event reconstruction improved the spatial resolution from 164.2 to 91.9 $μ$m at 10% modulation transfer function (MTF), while a 2.1$\times$ configuration with a $\mathrm{Gd_2O_2S:Tb}$ scintillator achieved 26.1 $μ$m at the same MTF threshold. Following event reconstruction and rejection of single-photon reconstructed events, the mean peak-normalized spectral response in the 6-8 angstrom out-of-band interval decreased by 85.4% relative to the raw-pixel wavelength-count spectrum, indicating reduced afterglow-related spectral background. Under the tested pulsed-beam conditions, the maximum instantaneous pixel-hit rate per pixel was approximately four times its time-averaged value, providing an experimental basis for assessing readout requirements at higher neutron flux. Conventional radiography resolved structural details, while energy-selective imaging enhanced the contrast between materials with similar broadband transmission and Bragg-edge imaging revealed crystallographic features in steel. These results demonstrate the imaging-quality benefits and capabilities of scintillator-based event-mode neutron imaging.

physics.ins-det↗

Action-Conditioned World Model for Goal Plane Guidance in Robotic Neck Ultrasound

Ultrasound examinations typically follow specific scanning protocols that require navigating through an ordered sequence of anatomical landmarks and target imaging planes. We present a local action-conditioned world model framework for goal plane probe guidance in robotic ultrasound. Autonomous ultrasound tasks often require large numbers of probe-motion trajectories for training, but collecting high-quality demonstrations is labor-intensive and explicit simulators are difficult to build because ultrasound appearance depends on contact, tissue deformation, and view-dependent acoustic artifacts. We address this problem with a two-stage model-based learning pipeline. First, a latent conditional diffusion world model predicts future ultrasound observations from recent context frames, probe motions and temporal offset. Second, a goal-conditioned temporal transformer predicts ordered probe motions and is fine-tuned using rewards from the frozen world model. Experiments on the self-collected dataset show that the world model preserves action-dependent anatomical structure on target-directed scans. In real-world closed loop experiments, the framework achieves success rates of 70.0\% for carotid guidance and 65.0\% for thyroid guidance. These results demonstrate the potential of learned ultrasound dynamics for training goal-directed robotic probe navigation within protocol-constrained organ specific tasks.

cs.RO↗

Denser $\neq$ Better: Limits of On-Policy Self-Distillation for Continual Post-Training

Continual post-training enables foundation models to acquire new knowledge while preserving existing capabilities. Recent work suggests that on-policy learning can mitigate forgetting, with self-distillation as a particularly attractive approach. We revisit this optimistic claim through self-distillation policy optimization (SDPO). Our experiments show that SDPO accelerates in-domain specialization when teacher signals are stable and well aligned, but struggles to generalize out of distribution. In continual post-training, SDPO exhibits greater forgetting and can even collapse, whereas GRPO, the more established on-policy reinforcement learning method, adapts more conservatively and better preserves prior capabilities. Further analyses link these failures to increased drift in parameter and response space, and to amplification of high-frequency artifacts through a self-reinforcing teacher-student loop. Thus, on-policy data alone is insufficient for continual learning. Self-distillation is effective when teacher targets are stable and token-level supervision is reliable, but should not be treated as a default stabilizer for continual post-training. Our code is available at https://github.com/Moenupa/SDPO-CL.

cs.LG↗

SemPSG: A Semantic Channel-Aware Foundation Model for Polysomnography Analysis

Polysomnography (PSG) integrates multiple physiological signals to provide a comprehensive characterization of human sleep, yet its heterogeneous channel configurations across centers pose substantial challenges for transferable representation learning. Existing foundation models mainly focus on physiological modeling or temporal learning, while channel identity is often treated as a fixed structural index, overlooking the physiological semantics encoded by signal modality and reference configuration. To this end, we propose SemPSG, a Semantic channel-aware foundation model for heterogeneous PSG analysis. SemPSG explicitly represents the physiological semantics of channel identity and incorporates them into both signal representation learning and channel aggregation, enabling flexible modeling across diverse data configurations. Specifically, a semantic-conditioned time-series encoder captures signal-specific temporal dynamics and cross-signal interactions, while a multi-view image encoder extracts complementary time-frequency and morphological patterns from the same physiological recordings. We evaluate SemPSG on sleep and health-related tasks, including sleep staging, sleep-disorder breathing analysis, disease prediction, cognition and emotion recognition, and demographic estimation. Extensive experiments demonstrate consistent improvements over both general-purpose time series foundation models and PSG-specific foundation models, together with generalization across heterogeneous datasets across diverse channel configurations.

cs.LG↗

Med-RADIO: Reducing All Medical Domains Into One via Multi-Teacher Distillation

The rapid expansion of large-scale medical datasets and computational resources has driven significant progress in medical foundation models. Given the inherent heterogeneity of medical imaging modalities, current research mainly follows two paths: specialized models optimized for specific modalities, and generalist models designed to handle multiple modalities. However, medical generalist models suffer from both insufficient training data scale relative to natural image generalists and inadequate domain-specific depth relative to medical specialists. Empirically, generalist models establish a cross-modality performance baseline, while specialists define the performance ceiling within their respective domains. To elevate this baseline toward these ceilings, we propose Med-RADIO, a medical multi-teacher distillation framework that Reduces All Domains Into One by compressing complementary expertise from multiple domain-specific teachers into a unified medical vision foundation model. Our method curates both generalist and specialist teachers, allocates modality-aligned distillation streams to reorganize generalist pretraining data so it matches specialist domains, and uses a balanced loss to prevent any single teacher from dominating the distillation process. On internal and external classification benchmarks spanning five modalities, Med-RADIO improves over strong medical generalists under linear probing and remains competitive with representative specialists on most evaluated modalities. Code is available at https://github.com/CAIR-HKISI/Med-RADIO.

cs.CV↗

Quantum Feature Selection for Biomedical Data Analysis

Feature selection is an essential step for reducing complexity of high dimensional data, usually in preparation for developing machine learning models such as computational biomarkers. However, there are limitations associated with feature selection such as for metabolomic data where there are hundreds or even thousands of features per study participant, while the available number of participants in a clinical trials is limited. Classical methods such as exhaustive search require evaluation of all possible feature combinations, making them costly in terms of computation and runtime, or even infeasible, as feature dimensionality increases. In this study, we propose a novel Quadratic Unconstrained Binary Optimization (QUBO) coefficient formulation and pose metabolomic feature selection as a QUBO problem that selects a specified number of features by balancing their relevance against redundancy among the selected variables. To evaluate our proposed QUBO objective function, we conducted a series of experiments using Bias-Field Digitized Counterdiabatic Quantum Optimization (BF-DCQO) and Quantum Approximate Optimization Algorithm (QAOA) on a quantum gate based computer. We also compared the method to several classical methods on three metabolomic datasets associated with Autism Spectrum Disorder (ASD) on a classical computer. Our method reduces runtime compared with exhaustive search and Iterative Tabu Search (ITS) and achieves competitive performance across classifiers compared to classical filter, wrapper, and embedded methods. These results do not claim quantum advantage; rather, they establish hardware feasibility and demonstrate the current capabilities of Noisy Intermediate-Scale Quantum (NISQ) devices.

quant-ph↗

SurgMotion: A Video-Native Foundation Model for Universal Understanding of Surgical Videos

While foundation models have advanced surgical video analysis, current approaches rely predominantly on pixel-level reconstruction objectives that waste model capacity on low-level visual details, such as smoke, specular reflections, and fluid motion, rather than semantic structures essential for surgical understanding. We present SurgMotion, a video-native foundation model that shifts the learning paradigm from pixel-level reconstruction to latent motion prediction. Built on the Video Joint Embedding Predictive Architecture (V-JEPA), SurgMotion introduces three key technical innovations tailored to surgical videos: (1) motion-guided latent masked prediction to prioritize semantically meaningful regions, (2) spatiotemporal affinity self-distillation to enforce relational consistency, and (3) spatiotemporal feature diversity regularization (SFDR) to prevent representation collapse in texture-sparse surgical scenes. To enable large-scale pretraining, we curate SurgMotion-15M, the largest surgical video dataset to date, comprising 3,658 hours of video from 50 sources across 13 anatomical regions. Extensive experiments across 17 benchmarks demonstrate that SurgMotion significantly outperforms state-of-the-art methods on surgical workflow recognition, achieving 14.6 percent improvement in F1 score on EgoSurgery and 10.3 percent on PitVis; on action triplet recognition with 39.54 percent mAP-IVT on CholecT50; as well as on skill assessment, polyp segmentation, and depth estimation. These results establish SurgMotion as a new standard for universal, motion-oriented surgical video understanding.

cs.CV↗

Clifford-efficient sparse state preparation for molecular wavefunctions

Sparse quantum state preparation concerns an $n$-qubit target state that is a superposition of only $d \ll 2^n$ computational basis states. Existing approaches exploit this sparsity by compressing these $d$ basis states and their amplitudes onto a smaller set of qubits, called the dense register, before expanding the prepared state to the full register. Rather than relying on the permutation-based compression used in prior work, we exploit affine relationships among the binary configurations over the finite field $\operatorname{GF}(2)$ to reduce both the non-Clifford gate count and the ancillary qubit count. Invertible affine transformations over $\operatorname{GF}(2)$, comprising Gaussian elimination and all-ones-row removal, first reduce the dense register from $n$ to the rank $r$ using only Clifford gates and no ancillary qubits. An optional binary encoding stage then trades additional Toffoli gates and ancillary qubits for further compression to the minimum $\lceil\log_2 d\rceil$ dense qubits needed to represent $d$ distinct configurations. For chemically relevant wavefunctions, such as those obtained from selected configuration interaction calculations, shared electronic excitation patterns produce many of these affine relationships, enabling substantial Clifford-only compression before binary encoding. Across the molecular benchmarks, our method requires the fewest ancillary qubits among the evaluated sparse state preparation methods while maintaining comparable non-Clifford gate counts when using binary encoding.

quant-ph↗

Geometry-Aware Camera Localization for Bronchoscopy

Camera localization in bronchoscopy remains a challenging problem due to stringent accuracy requirements, real-time constraints, and limited training data. Compared to natural scenes, the confined anatomical structures demand millimeter-level precision, while intraoperative guidance necessitates low-latency inference. However, existing methods often fail to effectively exploit preoperative geometric priors, limiting their robustness and accuracy. To address these limitations, we propose a unified geometry-aware bronchoscope localization framework (GABL) that effectively fuses preoperative structural priors with paired intraoperative video to estimate 6-DoF camera poses. Specifically, to address visual ambiguity in complex airways, we propose a graph-guided coarse-to-fine localization scheme that effectively leverages structural priors for precise pose estimation. Furthermore, to mitigate pose jitter and bridge the visual-structural gap, we integrate a Transformer-based tracking model with a novel RGB-depth matching objective, jointly enforcing spatio-temporal and geometric consistency. Extensive experiments demonstrate that our method yields remarkable reductions of 8.37% and 31.76% in translation and rotation errors over the prior state-of-the-art, alongside 4 times inference speedup (33.6 FPS) for robust real-time bronchoscope localization. Project website: https://paulili08.github.io/GABL/.

cs.CV↗

Learning-based Hierarchical Tracheal Anatomy Understanding from Sparse Surgical Demonstration Annotations for Ultrasound Robots

Tracheostomy requires precise localization of the tracheal incision site; however, conventional manual palpation is subjective and often unreliable, while ultrasound utility remains operator-dependent. This work presents a learning-based framework for hierarchical tracheal anatomy understanding, designed specifically for ultrasound-guided robotic systems. We propose a two-stage perception pipeline integrating a YOLOv8n localization backbone with a sparse, prompt-optimized SAM2 decoder to achieve high-fidelity segmentation from sparse surgical annotations. Our hybrid training strategy, bridging curated laboratory data with unconstrained sequences, ensures clinical robustness. Experimental benchmarks demonstrate that this decoupled architecture effectively balances generalization, precision, and efficiency. The YOLOv8n and SAM2 framework achieves a consistent Mean Dice Similarity Coefficient (DSC) of 0.777 across both controlled and generalized domains. This significantly outperforms U-Net baselines, which often suffer from anatomical fragmentation and performance degradation (Generalization DSC $\le$ 0.494). By constraining mask decoding to targeted, sparse regions of interest, our model achieves a throughput of 6.92 FPS, which is vital for closed-loop robotic teleoperation. This study confirms that a robust hierarchical understanding of tracheal anatomy can be derived by coupling lightweight localization with foundation-scale visual models. Our framework establishes a scalable foundation for standardized, autonomous surgical assistance, effectively navigating the variability of real-world ultrasound to enhance the safety and precision of robotic-assisted tracheostomy.

eess.IV↗

Reinforcement Fine-Tuning Naturally Mitigates Forgetting in Continual Post-Training

Continual post-training (CPT) is a popular and effective technique for adapting foundation models like multimodal large language models to ever-evolving downstream tasks. While existing research primarily focuses on methods like data replay, model expansion, or parameter regularization, the fundamental role of the learning paradigm remains largely unexplored. This paper presents a comparative analysis of two core post-training paradigms: supervised fine-tuning (SFT) and reinforcement fine-tuning (RFT), investigating their respective impacts on knowledge retention during CPT. Our experiments are conducted across multiple multimodal tasks, utilizing Qwen2.5-VL-7B-Instruct as the base model. The investigation yields two significant findings: (1) When continuously learning on downstream tasks, SFT leads to catastrophic forgetting of previously learned tasks. In contrast, RFT inherently preserves prior knowledge and achieves performance comparable to multi-task training. (2) RFT successfully protects and even enhances the model's general knowledge on standard benchmarks, while SFT degrades general model capabilities severely. Further analysis reveals that this stability is not primarily due to explicit mechanisms like KL penalty or chain-of-thought reasoning. We investigate RFT's learning dynamics and find that its selective update mechanism inherently prevents interference with established knowledge. Based on this insight, we propose a rollout-based instance filtering algorithm (RIF-RFT) that enhances the training efficiency of RFT by focusing on learnable samples. Our comprehensive study demonstrates the superiority of RFT as a robust paradigm for continual post-training.

cs.LG↗

Bridging the Intention-Expression Gap: Aligning Multi-Dimensional Preferences via Hierarchical Relevance Feedback in Text-to-Image Diffusion

Users often possess a clear visual intent but struggle to articulate it precisely in language. This intention-expression gap makes aligning generated images with latent visual preferences a fundamental challenge in text-to-image diffusion models. Existing methods either require model training, sacrificing flexibility, or rely on textual feedback, imposing a heavy cognitive burden. Although recent training-free methods use click-based binary preference feedback to reduce user effort, they force Foundation Models (FMs) to infer preferences at the semantic level. When faced with multi-dimensional preferences, FMs suffer from inference overload and fail to identify exact preferred feature values under conflicting user signals. Consequently, a flexible framework for multi-dimensional feature alignment remains absent. To address this, we propose a Hierarchical Relevance Feedback-Driven (HRFD) framework. Recognizing that multiple features struggle to converge simultaneously, HRFD organizes them into a three-tier hierarchy and adapts relevance feedback to enforce coarse-to-fine convergence, minimizing cognitive load. To bypass FM inference overload, HRFD decouples the process into independent single-feature preference inference tasks. Furthermore, to overcome FMs' failure in identifying preferred values, HRFD employs statistical inference to quantify the distribution divergence of features between "liked" and "disliked" image sets, achieving robust and transparent preference measurement. Crucially, HRFD operates entirely within the external text space, remaining strictly training-free and model-agnostic. Extensive experiments demonstrate that HRFD effectively captures the user's true visual intent, significantly outperforming baseline approaches.

cs.CV↗

Multimodal Causal-Driven Representation Learning for Generalizable Medical Image Segmentation

Vision-Language Models (VLMs), such as CLIP, have demonstrated remarkable zero-shot capabilities in various computer vision tasks. However, their application to medical imaging remains challenging due to the high variability and complexity of medical data. Specifically, medical images often exhibit significant domain shifts caused by various confounders, including equipment differences, procedure artifacts, and imaging modes, which can lead to poor generalization when models are applied to unseen domains. To address this limitation, we propose Multimodal Causal-Driven Representation Learning (MCDRL), a novel framework that integrates causal inference with the VLM to tackle domain generalization in medical image segmentation. MCDRL is implemented in two steps: first, it leverages CLIP's cross-modal capabilities to identify candidate lesion regions and construct a confounder dictionary through text prompts, specifically designed to represent domain-specific variations; second, it trains a causal intervention network that utilizes this dictionary to identify and eliminate the influence of these domain-specific variations while preserving the anatomical structural information critical for segmentation tasks. Extensive experiments demonstrate that MCDRL consistently outperforms competing methods, yielding superior segmentation accuracy and exhibiting robust generalizability.

cs.CV↗

Reward-Guided Semantic Evolution for Test-time Adaptive Object Detection

Open-vocabulary object detection with vision-language models (VLMs) such as Grounding DINO suffers from performance degradation under test-time distribution shifts, primarily due to semantic misalignment between text embeddings and shifted visual embeddings of region proposals. While recent test-time adaptive object detection methods for VLM-based either rely on costly backpropagation or bypass semantic misalignment via external memory, none directly and efficiently align text and vision in a training-free manner. To address this, we propose Reward-Guided Semantic Evolution (RGSE), a training-free framework that directly refines the text embeddings at test time. Inspired by evolutionary search, RGSE treats text embedding adaptation as a semantic search process: it perturbs text embeddings as candidate variants, evaluates them via cosine similarity with current and historical high-confidence visual proposals as a reward signal, and fuses them into a refined embedding through reward-weighted averaging. Without any backpropagation, RGSE achieves state-of-the-art performance across multiple detection benchmarks while adding minimal computational overhead. Our code will be open source upon publication.

cs.CV↗

Robustness of Vision Foundation Models to Common Perturbations

A vision foundation model outputs an embedding vector for an image, which can be affected by common editing operations (e.g., JPEG compression, brightness, contrast adjustments). These common perturbations alter embedding vectors and may impact the performance of downstream tasks using these embeddings. In this work, we present the first systematic study on foundation models' robustness to such perturbations. We propose three robustness metrics and formulate five desired mathematical properties for these metrics, analyzing which properties they satisfy or violate. Using these metrics, we evaluate six industry-scale foundation models (OpenAI, Meta) across nine common perturbation categories, finding them generally non-robust. We also show that common perturbations degrade downstream application performance (e.g., classification accuracy) and that robustness values can predict performance impacts. Finally, we propose a fine-tuning approach to improve robustness without sacrificing utility.

cs.CR↗

Leave My Images Alone: Preventing Multi-Modal Large Language Models from Analyzing Images via Visual Prompt Injection

Multi-modal large language models (MLLMs) have emerged as powerful tools for analyzing Internet-scale image data, offering significant benefits but also raising critical safety and societal concerns. In particular, open-weight MLLMs may be misused to extract sensitive information from personal images at scale, such as identities, locations, or other private details. In this work, we propose ImageProtector, a user-side method that proactively protects images before sharing by embedding a carefully crafted, nearly imperceptible perturbation that acts as a visual prompt injection attack on MLLMs. As a result, when an adversary analyzes a protected image with an MLLM, the MLLM is consistently induced to generate a refusal response such as "I'm sorry, I can't help with that request." We empirically demonstrate the effectiveness of ImageProtector across six MLLMs and four datasets. Additionally, we evaluate three potential countermeasures, Gaussian noise, DiffPure, and adversarial training, and show that while they partially mitigate the impact of ImageProtector, they simultaneously degrade model accuracy and/or efficiency. Our study focuses on the practically important setting of open-weight MLLMs and large-scale automated image analysis, and highlights both the promise and the limitations of perturbation-based privacy protection.

cs.CV↗

Integration and characterization of Readout Electronics System for dN/dx Measurement with Drift Chamber Prototype

To explore the feasibility of high-precision particle identification using the cluster counting technique for the drift chamber, a dedicated readout electronics system with low noise, high bandwidth, and high sampling rate is required. This paper presents the design and performance evaluation of a scalable readout prototype developed for this application. The system architecture integrates a custom front-end with a $1.3\ \text{GSps}$ waveform sampling backend, implemented within a modular 120-channel framework. Laboratory characterization of the 40-channel prototype demonstrates a $-3$ dB analog bandwidth of $460\ \text{MHz}$ and an Equivalent Noise Input current of $0.81\ μ\text{A}_\text{rms}$. These specifications are essential for preserving the fast temporal features of ionization signals. Furthermore, the system achieves an intrinsic timing jitter of $0.87\ \text{ns}$, which satisfies the timing precision requirements for drift distance measurement. Joint experiments with a drift chamber prototype using cosmic rays verified the system's capability to resolve discrete ionization peaks within piled-up waveforms. These results confirm that the readout electronics provide the signal fidelity and temporal resolution necessary for future cluster counting algorithm development.

physics.ins-det↗

Steering LLMs for Culturally Localized Generation

LLMs are deployed globally, yet produce responses biased towards cultures with abundant training data. Existing cultural localization approaches such as prompting or post-training alignment are black-box, hard to control, and do not reveal whether failures reflect missing knowledge or poor elicitation. In this paper, we address these gaps using mechanistic interpretability to uncover and manipulate cultural representations in LLMs. Leveraging sparse autoencoders, we identify interpretable features that encode culturally salient information and aggregate them into Cultural Embeddings (CuE). We use CuE both to analyze implicit cultural biases under underspecified prompts and to construct white-box steering interventions. Across multiple models, we show that CuE-based steering increases cultural faithfulness and elicits significantly rarer, long-tail cultural concepts than prompting alone. Notably, CuE-based steering is complementary to black-box localization methods, offering gains when applied on top of prompt-augmented inputs. This also suggests that models do benefit from better elicitation strategies, and don't necessarily lack long-tail knowledge representation, though this varies across cultures. Our results provide both diagnostic insight into cultural representations in LLMs and a controllable method to steer towards desired cultures.

cs.CL↗