Searcharxiv⌕ Search

arXiv subjects

Yizhi Wang

Publications and source records attributed to Yizhi Wang.

At least 19 recordsLinked to original sources
arXiv2026

Truncated hybrid tensor networks for distributed quantum simulation

Simulation of quantum many-body systems is a principal application of quantum computing, but available devices remain limited by the number of qubits and cannot accommodate systems of the desired size. One approach is to host a large evolution across several modest distributed systems that exchange classical information alone. Current protocols such as circuit knitting incur a quasi-probability sampling cost that grows exponentially with the number of remote gates across the cut, even when physical correlations across the cut are actually bounded. In this work, we present a truncated hybrid tensor network (THTN) framework in which a remote two-body gate is applied as a sum of local unitaries on the subsystems, linked by a classical connecting tensor, and an interface truncation retains Schmidt modes across the cut. Remote gates between the same subsystems can be merged on one connector, so non-one-dimensional models, such as layered partitions with strong intra-layer and weaker interlayer couplings, may be cast onto a one-dimensional cut. The retained non-negative Schmidt coefficients also define a sampling rule for local observables. We validate this distributed protocol by classical simulation against time-evolving block decimation (TEBD) at the same bond dimension. The truncated dynamics track the TEBD references on chains and ladders, with the clearest gain on a layered model.

quant-ph↗
arXiv2026

ConRub-Med: Reinforcement Learning with Consensus Rubrics for Open-Ended Medical Question Answering

Reinforcement learning with verifiable rewards has been especially effective in mathematics and coding, where answers can be checked automatically. Many open-ended medical questions lack comparably cheap outcome verifiers: responses may be partly correct, incomplete, or contain clinically consequential errors. Rubrics written or validated by physicians offer strong clinical grounding, but involving experts in every instance is costly. Model-generated rubrics make this supervision scalable. We introduce ConRub-Med to preserve useful distinctions as rubric feedback moves from construction to policy optimization. For each prompt, three heterogeneous language models propose atomic criteria independently; a separate model reviews them, retaining only criteria with semantic support from all three generators. Three-State scoring distinguishes correct coverage, missing information, and incorrect claims. Errors receive negative rather than zero credit. When every response in a complete Group Relative Policy Optimization (GRPO) group receives the same final reward, a pairwise judge provides sequence advantages only if both candidate orders agree, without changing the scalar rewards. Groups without ties use vanilla GRPO. In a blinded study matched by question, two medical experts rate panels from the full pipeline as more clinically relevant than panels produced by one generator. Across the evaluated open models, ConRub-Med ranks first on six of nine benchmarks and achieves the highest medical and generalization averages. Using the resulting rubric dataset of 5,166 prompts, it scores $38.98 \pm 1.04$ (mean $\pm$ SD) on HealthBench-Hard, compared with InfiMed-ORBIT's 33.60 with 8,000 samples and 37.30 with 28,000.

cs.CL↗
arXiv2026

Expressive Power and Limitations of Multi-photon Quantum Neural Networks

Quantum neural networks (QNNs) have shown promise in leveraging quantum computation for machine learning tasks. Utilizing multiple identical photons as input, multi-photon quantum neural networks (MPQNNs) have the potential to enhance the expressivity through increasing the photon number. However, how precisely the expressivity of an MPQNN is affected by an increase in photon number, and whether it can be infinitely enhanced by increasing the photon number, remains unexplored. In this work, we quantitatively estimate the expressivity of this model by deriving upper bounds on approximation error in two cases. In the case of a fixed observable, there exists a threshold that scales linearly with the mode number. Below the threshold, the expressivity of an MPQNN can be enhanced polynomially by increasing the photon number. Above the threshold, however, increasing the photon number does not affect the expressivity. In the case of a trainable observable, the expressivity can always be enhanced polynomially by increasing the photon number. These findings are then validated by numerical simulations. Our work elucidates the performance enhancement of multi-photon quantum feature in QNNs, as well as its limitations, offering guidance for leveraging multi-photon advantages in quantum machine learning.

quant-ph↗
arXiv2026

SmartHomeSecure: Automated Detection and Repair of Smart Home Configuration Errors Using Large Language Models

Smart home automation platforms increasingly rely on user-authored YAML configuration files to define device behaviors, but these files are prone to syntax, formatting, and semantic logic errors that can cause automation failures and safety risks. Existing YAML validators, static analysis tools, and general-purpose large language models offer limited support for end-to-end diagnosis and repair because they lack domain-specific understanding and validated correction workflows. This paper presents SmartHomeSecure, a prototype for automated detection and repair of Home Assistant configuration errors using lightweight program analysis and constraint-guided large language model generation. SmartHomeSecure parses YAML files, detects syntactic and common semantic errors, normalizes error context, applies deterministic auto-fixes for routine defects, and constructs constrained prompts that guide LLMs toward minimal and structurally valid repairs. The system is implemented as a modular web application with four layers: UI Shell, Feature Orchestrator, Domain Engine, and Integration Layer. Its repair pipeline was evaluated on 100 real-world Home Assistant YAML files with manually injected errors across five categories: syntax/parsing, indentation, mapping, sequence, and scalar quoting errors. Four models were tested: gpt-oss-20b, gpt-oss-120b, llama-3.1-8b, and llama-3.3-70b. Results show that three models achieved 100% error detection accuracy, with repair success rates ranging from 87% to 93%. Manual verification found no hallucinated or incorrect repairs among successful outputs. These findings suggest that combining domain-aware program analysis with constrained generative AI is a feasible approach for improving the reliability and usability of smart home configuration repair.

cs.SE↗
arXiv2026

Teaching Vision-Language-Action Models What to See and Where to Look

Vision-Language-Action (VLA) models have emerged as a promising paradigm for end-to-end autonomous driving. However, existing VLAs' training relies heavily on text-centric visual question answering and chain-of-thought reasoning data, which emphasizes linguistic reasoning rather than action-grounded planning. As a result, the learned representations capture semantic knowledge but lack spatial dependencies crucial for reliable trajectory prediction. We propose DriveTeach-VLA, a framework that explicitly teaches VLAs what to see and where to look. Driving-aware Vision Distillation (DVD) injects driving-specific perceptual priors into the vision encoder, while 2D Trajectory-Guided Prompts (2D-TGP) provide spatial conditioning aligned with feasible driving trajectories. Together, they form a vision-guided learning pipeline: what to see (DVD pretraining) - where to look (TGP-guided SFT) - how to act (TGP-guided GRPO). DriveTeach-VLA achieves the state-of-the-art performance on NAVSIM and nuScenes. Our code is available at: https://github.com/ShivaTeam/DriveTeach-VLA.

cs.CV↗
arXiv2026

Transversality Methods for Homotopy Groups of Stable Loci in Affine GIT Quotients

We investigate the homotopy groups of stable loci in affine Geometric Invariant Theory (GIT), arising from linear actions of complex reductive algebraic groups on complex affine spaces. Our approach extends the infinite-dimensional transversality framework of Daskalopoulos-Uhlenbeck and Wilkin to this general GIT setting. Central to our method is the construction of a G-equivariant holomorphic vector bundle over the conjugation orbit of a one-parameter subgroup (1-PS), whose fibres are precisely the negative weight spaces determining instability. A key proposition establishes that a naturally defined evaluation map is transverse to the zero section of this bundle, implying that generic homotopies avoid all unstable and strictly semistable strata under certain dimensional inequalities. Our result also covers cases where semistability does not coincide with stability. The applicability of this framework is illustrated by several examples. In linear control theory, where GIT stability corresponds to the notion of controllability, our results determine the connectivity of the space of controllable systems. In statistical modelling, where stability for star-shaped Gaussian model corresponds to the existence of a unique Maximum Likelihood Estimate, we compute the connectivity of the space of data samples that yield such a unique estimate, providing topological insight into the problem of parameter non-identifiability. We also consider Helmke systems and show that for stability parameters satisfying certain bounds, the space of systems that are both controllable and observable is exactly the space of stable points. The main result can then be used to compute the connectivity of this space.

math.AG↗
arXiv2026

On-Policy Replay for Continual Supervised Fine-Tuning

Continual supervised fine-tuning (SFT) is the de facto recipe for adapting large language models (LLMs) to a stream of downstream tasks, but it suffers from catastrophic forgetting of earlier capabilities. Recent work shows that on-policy signals -- training on the model's own outputs -- reduce forgetting more reliably than off-policy supervision. Existing on-policy methods route this signal through a new training objective (e.g., self-distillation losses with a teacher copy), inheriting an extra forward pass, schedule sensitivity, and stylistic drift from the teacher.We instead route the on-policy signal through the training data source. Our method, On-Policy Replay (OPR), rolls out the most recent checkpoint on a small budget of historical prompts, filters the generations by a task reward, and replays the surviving (prompt, model response) pairs as ordinary SFT examples. There is no teacher, no auxiliary loss, and no on-the-fly distillation. Across three 7--8B instruction-tuned backbones (Qwen2.5-7B-Instruct, Qwen3-8B, Llama3.1-8B-Instruct) on the TRACE continual-learning benchmark, OPR consistently reduces forgetting; on the sharpest stress test (Qwen2.5-7B-Instruct, Sequential SFT BWT -13.93), OPR lifts BWT to -0.65 at a 10% replay budget and to -2.29 at a 1% budget -- a 46% reduction in |BWT| over a tuned Vanilla Replay baseline, with 42--46% reductions observed across all three backbones. We give a KL-shrinkage interpretation that places OPR and prior on-policy distillation methods on a single axis, and we present a counterintuitive finding that explains why Vanilla Replay is already a strong baseline: low-score replay is uniformly worse than Vanilla Replay, demonstrating that the active ingredient in OPR is the on-policy distribution, not the response quality alone.Our code is available at https://github.com/Yancey2024/OnPolicyReplay.

cs.LG↗
arXiv2026

SlideCheck: Guiding Self-Supervised Pretraining of Pathology Foundation Models via Dataset Distributions

Pathology foundation models are pretrained on large streams of WSI-derived patches, while supervision during data construction is often slide-level, sparse, or heterogeneous. This mismatch makes it difficult to understand and control which biological patterns enter the pretraining data. We propose SlideCheck, a lightweight pretraining data guidance tool built on frozen pathology foundation model patch features. Rather than serving as a standalone patch diagnostic model, SlideCheck provides explicit abnormality and malignancy scores for organizing, filtering, and auditing pathology pretraining data. SlideCheck uses a dual-head MLP to separately model broad abnormal morphology and malignant evidence. A regularized feature-space scorer provides a supervised anchor for patch-level evidence estimation, while score-attention agreement combines patch scores with WSI-level MIL attention to mine high-confidence pseudo labels. The same scores are then used to construct broad-positive ViT pretraining subsets, where a patch is selected if either abnormality or malignancy evidence exceeds a threshold. Experiments show that SlideCheck-defined data distributions influence the downstream behavior of self-supervised ViT pretraining, indicating that biological composition is an important controllable factor in pathology foundation model development. Curated subsets can approach full-data performance, suggesting that explicitly scored patch pools may support more efficient and auditable pretraining data construction. These findings position SlideCheck as a data guidance and auditing layer for transforming large, undifferentiated patch pools into controllable and reusable pretraining datasets.

cs.CV↗
arXiv2026

BioArc: Discovering Optimal Neural Architectures for Biological Foundation Models

Foundation models have revolutionized various fields such as natural language processing (NLP) and computer vision (CV). While efforts have been made to transfer the success of the foundation models in general AI domains to biology, existing works focus on directly adopting the existing foundation model architectures from general machine learning domains without a systematic design considering the unique physicochemical and structural properties of each biological data modality. This leads to suboptimal performance, as these repurposed architectures struggle to capture the long-range dependencies, sparse information, and complex underlying ``grammars'' inherent to biological data. To address this gap, we introduce BioArc, a novel framework designed to move beyond intuition-driven architecture design towards principled, automated architecture discovery for biological foundation models. Leveraging Neural Architecture Search (NAS), BioArc systematically explores a vast architecture design space, evaluating architectures across multiple biological modalities while rigorously analyzing the interplay between architecture, tokenization, and training strategies. This large-scale analysis identifies novel, high-performance architectures, allowing us to distill a set of empirical design principles to guide future model development. Furthermore, to make the best of this set of discovered principled architectures, we propose and compare several architecture prediction methods that effectively and efficiently predict optimal architectures for new biological tasks. Overall, our work provides a foundational resource and a principled methodology to guide the creation of the next generation of task-specific and foundation models for biology.

cs.LG↗
arXiv2026

Is Class Signal Clustered or Routed in Task-Induced Implicit Neural Representation Weight Spaces?

Implicit neural representations (INRs) encode images as neural-network weights, making image classification a problem of weight-space classifiability. A natural geometric hypothesis is that classifier feedback should make image-specific weights cluster by class in the shared-anchor coordinate. We test this hypothesis in the SIREN-based Meta Weight Transformer (MWT) regime, where end-to-end training meta-learns a shared initialization and inner-loop update schedule for fitting image-specific SIRENs. We find that this prediction fails. Exposed weight-space geometry and supervised clustering pressure do not reliably track trained-reader accuracy; clustering can even make local neighborhoods more class-consistent while making the trained reader worse. Crucially, the reader constructs rather than inherits class-aligned geometry: token-flow diagnostics show that class-aligned neighborhoods become strongly predictive of trained-reader accuracy only after late reader interactions, not in the input coordinate. We further identify the native SIREN bias column in the augmented weight token as a low-dimensional, sample-dependent causal readout route for the trained reader; targeted controls rule out generic scalar-column and marginal-distribution artifacts. The diagnosis motivates interventions that strengthen reader routing, add an explicit bias route, or use denser inner-loop fitting; under the lane-specific training conventions used here, route-directed variants often outperform the shared-anchor baseline but interact non-additively. Task-induced INR weights are classifiable not because they form raw geometric clusters, but because their class signal is routed through the reader.

cs.CV↗
arXiv2026

High-efficiency Weak-trace-free Counterfactual Communication via Quantum Zeno Effect

The quantum Zeno effect, which inhibits quantum state evolution via repeated weak measurements, significantly enhances the efficiency of interaction-free measurement (IFM). This fundamental mechanism facilitates high-efficiency counterfactual quantum communication, enabling information delivery without particle transmission through the channel. However, the transmission time of the counterfactual communication requires minutes for bit and suffers the bit error when transmitting an image. Applying the quantum Zeno effect, we experimentally demonstrate high-efficiency weak-trace-free counterfactual communication on a quantum photonic chip, achieving a transmission probability of 74.2 $\pm$ 1.6\% for bit 0 and 85.1 $\pm$ 1.3\% for bit 1. Furthermore, we successfully transmit our group's logo -- Quanta -- through counterfactual communication, and reduce the time cost from minutes to seconds for bit, with zero bit errors after information processing. Our study provides a promising approach for secure and efficient communication using integrated silicon quantum photonics.

quant-ph↗
arXiv2026

Devil is in Narrow Policy: Unleashing Exploration in Driving VLA Models

We identify a fundamental Narrow Policy limitation undermining the performance of autonomous VLA models, where driving Imitation Learning (IL) tends to collapse exploration and limit the potential of subsequent Reinforcement Learning (RL) stages, which often saturate prematurely due to insufficient feedback diversity. Thereby, we propose Curious-VLA, a framework that alleviates the exploit-explore dilemma through a two-stage design. During IL, we introduce a Feasible Trajectory Expansion (FTE) strategy to generate multiple physically valid trajectories and a step-wise normalized trajectory representation to adapt this diverse data. In the RL stage, we present Adaptive Diversity-Aware Sampling (ADAS) that prioritizes high-diversity samples and introduce Spanning Driving Reward (SDR) with a focal style weighting to amplify reward's value span for improving sensitivity to driving quality. On the Navsim benchmark, Curious-VLA achieves SoTA results (PDMS 90.3, EPDMS 85.4) and a Best-of-N PDMS of 94.8, demonstrating its effectiveness in unlocking the exploratory potential of VLA models. Code: https://github.com/Mashiroln/curious_vla.git.

cs.CV↗
arXiv2026

GloPath: An Entity-Centric Foundation Model for Glomerular Lesion Assessment and Clinicopathological Insights

Glomerular pathology is central to the diagnosis and prognosis of renal diseases, yet the heterogeneity of glomerular morphology and fine-grained lesion patterns remain challenging for current AI approaches. We present GloPath, an entity-centric foundation model trained on over one million glomeruli extracted from 14,049 renal biopsy specimens using multi-scale and multi-view self-supervised learning. GloPath addresses two major challenges in nephropathology: glomerular lesion assessment and clinicopathological insights discovery. For lesion assessment, GloPath was benchmarked across three independent cohorts on 52 tasks, including lesion recognition, grading, few-shot classification, and cross-modality diagnosis-outperforming state-of-the-art methods in 42 tasks (80.8%). In the large-scale real-world study, it achieved an ROC-AUC of 91.51% for lesion recognition, demonstrating strong robustness in routine clinical settings. For clinicopathological insights, GloPath systematically revealed statistically significant associations between glomerular morphological parameters and clinical indicators across 224 morphology-clinical variable pairs, demonstrating its capacity to connect tissue-level pathology with patient-level outcomes. Together, these results position GloPath as a scalable and interpretable platform for glomerular lesion assessment and clinicopathological discovery, representing a step toward clinically translatable AI in renal pathology.

cs.CV↗
arXiv2026

Training Multimodal Large Reasoning Models Needs Better Thoughts: A Three-Stage Framework for Long Chain-of-Thought Synthesis and Selection

Large Reasoning Models (LRMs) have demonstrated remarkable performance on complex reasoning tasks through long Chain-of-Thought (CoT) reasoning. Extending these successes to multimodal reasoning remains challenging due to the increased complexity of integrating diverse input modalities and the scarcity of high-quality long CoT training data. Existing multimodal datasets and CoT synthesis methods still suffer from limited reasoning depth, modality conversion errors, and rigid generation pipelines, hindering model performance and stability. To this end, in this paper, we propose SynSelect, a novel three-stage Synthesis-Selection framework for generating high-quality long CoT data tailored to multimodal reasoning tasks. Specifically, SynSelect first leverages multiple heterogeneous multimodal LRMs to produce diverse candidate CoTs, and then applies both instance and batch level selection to filter high-quality CoTs that can effectively enhance the model's reasoning capabilities. Extensive experiments on multiple multimodal benchmarks demonstrate that models supervised fine-tuned on SynSelect-generated data significantly outperform baselines and achieve further improvements after reinforcement learning post-training. Our results validate SynSelect as an effective approach for advancing multimodal LRMs reasoning capabilities.

cs.AI↗
arXiv2026

Bridging Efficiency and Transparency: Explainable CoT Compression in Multimodal Large Reasoning Models

Long chains of thought (Long CoTs) are widely employed in multimodal reasoning models to tackle complex tasks by capturing detailed visual information. However, these Long CoTs are often excessively lengthy and contain redundant reasoning steps, which can hinder inference efficiency. Compressing these long CoTs is a natural solution, yet existing approaches face two major challenges: (1) they may compromise the integrity of visual-textual reasoning by removing essential alignment cues, and (2) the compression process lacks explainability, making it difficult to discern which information is critical. To address these problems, we propose XMCC, an eXplainable Multimodal CoT Compressor that formulates compression as a sequential decision-making process optimized via reinforcement learning. XMCC can effectively shorten reasoning trajectories while preserving key reasoning steps and answer correctness, and simultaneously generates natural-language explanations for its compression decisions. Extensive experiments on representative multimodal reasoning benchmarks demonstrate that XMCC not only reduces reasoning length but also provides explainable explanations, validating its effectiveness.

cs.AI↗
arXiv2026

Matrix Product State on a Quantum Computer

Solving quantum many-body systems is one of the most significant regimes where quantum computing applies. Currently, as a hardware-friendly computational paradigms, variational algorithms are often used for finding the ground energy of quantum many-body systems. However, running large-scale variational algorithms is challenging, because of the noise as well as the obstacle of barren plateaus. In this work, we propose the quantum version of matrix product state (qMPS), and develop variational quantum algorithms to prepare it in canonical forms, allowing to run the variational MPS method, which is equivalent to the Density Matrix Renormalization Group method, on near term quantum devices. Compared with widely used methods such as variational quantum eigensolver, this method can greatly reduce the number of qubits required, and thus can mitigate the effects of Barren Plateaus while obtain comparable or even better accuracy. Our method holds promise for distributed quantum computing, offering possibilities for fusion of different computing systems.

quant-ph↗
arXiv2026

StainNet: Scaling Self-Supervised Foundation Models on Immunohistochemistry and Special Stains for Computational Pathology

Foundation models trained with self-supervised learning (SSL) on large-scale histological images have significantly accelerated the development of computational pathology. These models can serve as backbones for region-of-interest (ROI) image analysis or patch-level feature extractors in whole-slide images (WSIs) based on multiple instance learning (MIL). Existing pathology foundation models (PFMs) are typically pre-trained on Hematoxylin-Eosin (H\&E) stained pathology images. However, images such as immunohistochemistry (IHC) and special stains are also frequently used in clinical practice. PFMs pre-trained mainly on H\&E-stained images may be limited in clinical applications involving these non-H\&E images. To address this issue, we propose StainNet, a collection of self-supervised foundation models specifically trained for IHC and special stains in pathology images based on the vision transformer (ViT) architecture. StainNet contains a ViT-Small and a ViT-Base model, both of which are trained using a self-distillation SSL approach on over 1.4 million patch images extracted from 20,231 publicly available IHC and special staining WSIs in the HISTAI database. To evaluate StainNet models, we conduct experiments on three in-house slide-level IHC classification tasks, three in-house ROI-level special stain and two public ROI-level IHC classification tasks to demonstrate their strong ability. We also perform ablation studies such as few-ratio learning and retrieval evaluations, and compare StainNet models with recent larger PFMs to further highlight their strengths. The StainNet model weights are available at https://github.com/WonderLandxD/StainNet.

cs.CV↗
arXiv2026

To What Extent Do Token-Level Representations from Pathology Foundation Models Improve Dense Prediction?

Pathology foundation models (PFMs) have rapidly advanced and are becoming a common backbone for downstream clinical tasks, offering strong transferability across tissues and institutions. However, for dense prediction (e.g., segmentation), practical deployment still lacks a clear, reproducible understanding of how different PFMs behave across datasets and how adaptation choices affect performance and stability. We present PFM-DenseBench, a large-scale benchmark for dense pathology prediction, evaluating 17 PFMs across 18 public segmentation datasets. Under a unified protocol, we systematically assess PFMs with multiple adaptation and fine-tuning strategies, and derive insightful, practice-oriented findings on when and why different PFMs and tuning choices succeed or fail across heterogeneous datasets. We release containers, configs, and dataset cards to enable reproducible evaluation and informed PFM selection for real-world dense pathology tasks. Project Website: https://m4a1tastegood.github.io/PFM-DenseBench

eess.IV↗