arXiv subjects

Zhiyi Chen

Publications and source records attributed to Zhiyi Chen.

At least 19 recordsLinked to original sources

arXiv2026

When Chatbots Accommodate: Auditing the Response Policies of AI Companions in Vulnerable Conversations

Millions turn to AI companion chatbots during loneliness, grief, and personal crises. How these companion platforms respond in such moments can shape the trajectory of a user's vulnerable state. Yet existing model audits evaluate reactions to pre-defined crisis prompts and miss the response policy that governs sustained real-world interaction. We address these gaps with two key contributions. First, we introduce the AI Companion Vulnerability-Response Taxonomy, a grounded, paired taxonomy of user vulnerability and chatbot response designed for analyzing extended companion chatbot interactions. Second, we apply Maximum Causal Entropy Inverse Reinforcement Learning to ~47k turns of real-world user conversations with GPT-4.1, Character.AI, and Replika to infer each platform's short-horizon response policy: the probability of each response category given the user's current vulnerability state. Our findings reveal distinct response profiles of AI companions in conversations with vulnerable users: GPT-4.1 reaches for advice, Character.AI spreads its response across different strategies, and Replika consistently asks questions and stays present. Over four weeks of repeated interaction, GPT-4.1 asks progressively fewer follow-up questions when users are distressed and increasingly sets boundaries or refers users out rather than pushing back. Within each platform, exploratory comparisons across user groups suggest that response policies also differ with users' pre-existing psychological risks and their bonds with the companion. Estimated model response policies are invisible to shallow behavioral audits, providing a new lens for auditing chatbots in the wild and enabling more realistic safety evaluation.

cs.HC↗

arXiv2026

Modulating task outcome value to mitigate real-world procrastination via noninvasive brain stimulation

Procrastination represents one of the most prevalent behavioral problems associated with individual health and societal productivity. Despite its high prevalence and substantial impact on daily functioning, its underlying neurocognitive mechanisms remain poorly understood. A leading model posits that procrastination arises from imbalanced competing motivations: the avoidance of negative task aversiveness and the pursuit of positive task outcomes, yet this framework has not been fully validated in real-world settings and not applied effectively to guide interventions. Here, we addressed this gap with a double-blind, randomized controlled trial. We applied seven sessions of high-definition transcranial direct current stimulation (HD-tDCS) to the left dorsolateral prefrontal cortex (DLPFC) in chronic procrastinators. Using the intensive experience sampling method (iESM), we assessed the effect of anodal HD-tDCS on real-world procrastination at offline after-effect (2-day interval) and long-term after-effect (6-month follow-up). We found that this neuromodulation produced a lasting reduction in real-world procrastination, with effects sustained at a 6-month follow-up. While the intervention is significantly associated with both decreased task aversiveness and increased perceived task outcome value, a mediation analysis indicated a disassociable mechanism: the increase in task outcome value (but not task aversiveness) showed a statistical pattern consistent with accounting for the observed behavioral improvement. In conclusion, the findings are consistent with the hypothesis that enhancing DLPFC function may reduce procrastination by selectively amplifying the valuation of future rewards, not by simply reducing negative feelings about the task. These results align with established decision-theoretic frameworks and suggest a targeted, theory-informed avenue for future behavioral interventions.

q-bio.NC↗

arXiv2026

FireRed-Image-Edit-1.0 Technical Report

We present FireRed-Image-Edit, a diffusion transformer for instruction-based image editing that achieves state-of-the-art performance through systematic optimization of data curation, training methodology, and evaluation design. We construct a 1.6B-sample training corpus, comprising 900M text-to-image and 700M image editing pairs from diverse sources. After rigorous cleaning, stratification, auto-labeling, and two-stage filtering, we retain over 100M high-quality samples balanced between generation and editing, ensuring strong semantic coverage and instruction alignment. Our multi-stage training pipeline progressively builds editing capability via pre-training, supervised fine-tuning, and reinforcement learning. To improve data efficiency, we introduce a Multi-Condition Aware Bucket Sampler for variable-resolution batching and Stochastic Instruction Alignment with dynamic prompt re-indexing. To stabilize optimization and enhance controllability, we propose Asymmetric Gradient Optimization for DPO, DiffusionNFT with layout-aware OCR rewards for text editing, and a differentiable Consistency Loss for identity preservation. We further establish REDEdit-Bench, a comprehensive benchmark spanning 15 editing categories, including newly introduced beautification and low-level enhancement tasks. Extensive experiments on REDEdit-Bench and public benchmarks (ImgEdit and GEdit) demonstrate competitive or superior performance against both open-source and proprietary systems. To support future research, our code, models, and benchmark suite are publicly available at https://github.com/FireRedTeam/FireRed-Image-Edit/ .

cs.CV↗

arXiv2026

TAHOE: Text-to-SQL with Automated Hint Optimization from Experience

Large Language Models (LLMs) have democratized database access through Text-to-SQL, but moving from prototypes to production remains difficult. Real deployments must handle strict SQL dialects, massive schemas, and evolving user preferences, while supervised fine-tuning is costly and rigid and agentic test-time scaling is expensive. We present Tahoe, a system that treats prompt optimization as a dynamic data management problem. Tahoe uses an error-driven hint learning pipeline across Development and Deployment to consolidate debugging traces into a structured Hint Bank. Compiler feedback is distilled into reusable Syntax Hints for dialect-specific rules, while execution and user feedback are converted into Semantic Hints for schema- and user-specific logic. Tahoe further introduces a Strategy Layer that models conflicting user intents as competing strategies under shared natural-language triggers, with recency signals and post-learning attribution statistics that summarize empirical success, harm, inertness, and support. At inference time, Tahoe retrieves relevant hints and guides the LLM through Logic Planning followed by SQL Synthesis. We implement and evaluate the development-phase workflow, leaving deployment-time human-feedback updates for future work. On Spider 2.0-Snow, Tahoe substantially improves Text-to-SQL without updating model parameters. On 113 supervised Spider 2.0-Snow-0212 examples using GPT-5.5, Tahoe raises pass rate from 61.95 percent to 79.42 percent and pass-at-4 from 72.57 percent to 87.61 percent, achieves 100 percent Snowflake syntax pass rate, and reduces average compiler-feedback critic rounds from 2.79 to 0.12 per sampled candidate. The same Hint Bank also transfers to weaker backbones, including a 19.7 percentage-point pass-rate gain on Doubao-2.0-lite.

cs.DB↗

arXiv2026

Gaussian Wardrobe: Compositional 3D Gaussian Avatars for Free-Form Virtual Try-On

We introduce Gaussian Wardrobe, a novel framework to digitalize compositional 3D neural avatars from multi-view videos. Existing methods for 3D neural avatars typically treat the human body and clothing as an inseparable entity. However, this paradigm fails to capture the dynamics of complex free-form garments and limits the reuse of clothing across different individuals. To overcome these problems, we develop a novel, compositional 3D Gaussian representation to build avatars from multiple layers of free-form garments. The core of our method is decomposing neural avatars into bodies and layers of shape-agnostic neural garments. To achieve this, our framework learns to disentangle each garment layer from multi-view videos and canonicalizes it into a shape-independent space. In experiments, our method models photorealistic avatars with high-fidelity dynamics, achieving new state-of-the-art performance on novel pose synthesis benchmarks. In addition, we demonstrate that the learned compositional garments contribute to a versatile digital wardrobe, enabling a practical virtual try-on application where clothing can be freely transferred to new subjects. Project page: https://ait.ethz.ch/gaussianwardrobe

cs.CV↗

arXiv2026

Growth First, Care Second? Tracing the Landscape of LLM Value Preferences in Everyday Dilemmas

People increasingly seek advice online from both human peers and large language model (LLM)-based chatbots. Such advice rarely involves identifying a single correct answer; instead, it typically requires navigating trade-offs among competing values. We aim to characterize how LLMs navigate value trade-offs across different advice-seeking contexts. First, we examine the value trade-off structure underlying advice seeking using a curated dataset from four advice-oriented subreddits. Using a bottom-up approach, we inductively construct a hierarchical value framework by aggregating fine-grained values extracted from individual advice options into higher-level value categories. We construct value co-occurrence networks to characterize how values co-occur within dilemmas and find substantial heterogeneity in value trade-off structures across advice-seeking contexts: a women-focused subreddit exhibits the highest network density, indicating more complex value conflicts; women's, men's, and friendship-related subreddits exhibit highly correlated value-conflict patterns centered on security-related tensions (security vs. respect/connection/commitment); by contrast, career advice forms a distinct structure where security frequently clashes with self-actualization and growth. We then evaluate LLM value preferences against these dilemmas and find that, across models and contexts, LLMs consistently prioritize values related to Exploration & Growth over Benevolence & Connection. This systemically skewed value orientation highlights a potential risk of value homogenization in AI-mediated advice, raising concerns about how such systems may shape decision-making and normative outcomes at scale.

cs.CY↗

arXiv2025

H-Neurons: On the Existence, Impact, and Origin of Hallucination-Associated Neurons in LLMs

Large language models (LLMs) frequently generate hallucinations -- plausible but factually incorrect outputs -- undermining their reliability. While prior work has examined hallucinations from macroscopic perspectives such as training data and objectives, the underlying neuron-level mechanisms remain largely unexplored. In this paper, we conduct a systematic investigation into hallucination-associated neurons (H-Neurons) in LLMs from three perspectives: identification, behavioral impact, and origins. Regarding their identification, we demonstrate that a remarkably sparse subset of neurons (less than $0.1\%$ of total neurons) can reliably predict hallucination occurrences, with strong generalization across diverse scenarios. In terms of behavioral impact, controlled interventions reveal that these neurons are causally linked to over-compliance behaviors. Concerning their origins, we trace these neurons back to the pre-trained base models and find that these neurons remain predictive for hallucination detection, indicating they emerge during pre-training. Our findings bridge macroscopic behavioral patterns with microscopic neural mechanisms, offering insights for developing more reliable LLMs.

cs.AI↗

arXiv2025

Synthetic Politics: Prevalence, Spreaders, and Emotional Reception of AI-Generated Political Images on X

Despite widespread concerns about the risks of AI-generated content (AIGC) to the integrity of social media discourse, little is known about its scale and scope, the actors responsible for its dissemination online, and the user responses it elicits. In this work, we measure and characterize the prevalence, spreaders, and emotional reception of AI-generated political images. Analyzing a large-scale dataset from Twitter/X related to the 2024 U.S. Presidential Election, we find that approximately 12% of shared images are detected as AI-generated, and around 10% of users are responsible for sharing 80% of AI-generated images. AIGC superspreaders--defined as the users who not only share a high volume of AI-generated images but also receive substantial engagement through retweets--are more likely to be X Premium subscribers, have a right-leaning orientation, and exhibit automated behavior. Their profiles contain a higher proportion of AI-generated images than non-superspreaders, and some engage in extreme levels of AIGC sharing. Moreover, superspreaders' AI image tweets elicit more positive and less toxic responses than their non-AI image tweets. This study serves as one of the first steps toward understanding the role generative AI plays in shaping online socio-political environments and offers implications for platform governance.

cs.SI↗

arXiv2025

Implementation of full-potential screened spherical wave based muffin-tin orbital for all-electron density functional theory

Screened spherical wave (SSW) of the Hankel function features the complete, minimal and short-ranged basis set, presenting a compact representation for electronic systems. In this work, we report the implementation of full-potential (FP) SSW based tight-binding linearized Muffin-Tin orbital (TB-LMTO) for all-electron density functional theory (DFT), and provide extensive tests on the robustness of FP-TB-LMTO and its high accuracy for first-principles material simulation. Through the introduction of double augmentation, SSW based MTO is accurately represented on the double grids including the full-space uniform and dense radial grids. Based on the the double augmentation, the accurate computation of full charge density, full potential,complex integral in the interstitial region and the total energy are all effectively addressed to realize the FP-TB-LMTO for DFT self-consistent calculations. By calculating the total energy,band structure, phase ordering, and elastic constants for a wide variety of materials, including normal metals, compounds, and diamond silicon, we domenstrate the highly accurate numerical implemetation of FP-TB-LMTO for all-electron DFT in comparison with other well-established FP method. The implementation of FP-TB-LMTO based DFT offers an important tool for the accurate first-principles tight-binding electronic structure calculations, particular important for the large-scale or strongly correlated materials.

cond-mat.mtrl-sci↗

arXiv2024

Deep Koopman-based Control of Quality Variation in Multistage Manufacturing Systems

This paper presents a modeling-control synthesis to address the quality control challenges in multistage manufacturing systems (MMSs). A new feedforward control scheme is developed to minimize the quality variations caused by process disturbances in MMSs. Notably, the control framework leverages a stochastic deep Koopman (SDK) model to capture the quality propagation mechanism in the MMSs, highlighted by its ability to transform the nonlinear propagation dynamics into a linear one. Two roll-to-roll case studies are presented to validate the proposed method and demonstrate its effectiveness. The overall method is suitable for nonlinear MMSs and does not require extensive expert knowledge.

eess.SY↗

arXiv2024

Co-Speech Gesture Video Generation via Motion-Decoupled Diffusion Model

Co-speech gestures, if presented in the lively form of videos, can achieve superior visual effects in human-machine interaction. While previous works mostly generate structural human skeletons, resulting in the omission of appearance information, we focus on the direct generation of audio-driven co-speech gesture videos in this work. There are two main challenges: 1) A suitable motion feature is needed to describe complex human movements with crucial appearance information. 2) Gestures and speech exhibit inherent dependencies and should be temporally aligned even of arbitrary length. To solve these problems, we present a novel motion-decoupled framework to generate co-speech gesture videos. Specifically, we first introduce a well-designed nonlinear TPS transformation to obtain latent motion features preserving essential appearance information. Then a transformer-based diffusion model is proposed to learn the temporal correlation between gestures and speech, and performs generation in the latent motion space, followed by an optimal motion selection module to produce long-term coherent and consistent gesture videos. For better visual perception, we further design a refinement network focusing on missing details of certain areas. Extensive experimental results show that our proposed framework significantly outperforms existing approaches in both motion and video-related evaluations. Our code, demos, and more resources are available at https://github.com/thuhcsi/S2G-MDDiffusion.

cs.CV↗

arXiv2023

Stochastic Deep Koopman Model for Quality Propagation Analysis in Multistage Manufacturing Systems

The modeling of multistage manufacturing systems (MMSs) has attracted increased attention from both academia and industry. Recent advancements in deep learning methods provide an opportunity to accomplish this task with reduced cost and expertise. This study introduces a stochastic deep Koopman (SDK) framework to model the complex behavior of MMSs. Specifically, we present a novel application of Koopman operators to propagate critical quality information extracted by variational autoencoders. Through this framework, we can effectively capture the general nonlinear evolution of product quality using a transferred linear representation, thus enhancing the interpretability of the data-driven model. To evaluate the performance of the SDK framework, we carried out a comparative study on an open-source dataset. The main findings of this paper are as follows. Our results indicate that SDK surpasses other popular data-driven models in accuracy when predicting stagewise product quality within the MMS. Furthermore, the unique linear propagation property in the stochastic latent space of SDK enables traceability for quality evolution throughout the process, thereby facilitating the design of root cause analysis schemes. Notably, the proposed framework requires minimal knowledge of the underlying physics of production lines. It serves as a virtual metrology tool that can be applied to various MMSs, contributing to the ultimate goal of Zero Defect Manufacturing.

cs.LG↗

arXiv2023

DiffPrep: Differentiable Data Preprocessing Pipeline Search for Learning over Tabular Data

Data preprocessing is a crucial step in the machine learning process that transforms raw data into a more usable format for downstream ML models. However, it can be costly and time-consuming, often requiring the expertise of domain experts. Existing automated machine learning (AutoML) frameworks claim to automate data preprocessing. However, they often use a restricted search space of data preprocessing pipelines which limits the potential performance gains, and they are often too slow as they require training the ML model multiple times. In this paper, we propose DiffPrep, a method that can automatically and efficiently search for a data preprocessing pipeline for a given tabular dataset and a differentiable ML model such that the performance of the ML model is maximized. We formalize the problem of data preprocessing pipeline search as a bi-level optimization problem. To solve this problem efficiently, we transform and relax the discrete, non-differential search space into a continuous and differentiable one, which allows us to perform the pipeline search using gradient descent with training the ML model only once. Our experiments show that DiffPrep achieves the best test accuracy on 15 out of the 18 real-world datasets evaluated and improves the model's test accuracy by up to 6.6 percentage points.

cs.DB↗

arXiv2021

Principal Gradient Direction and Confidence Reservoir Sampling for Continual Learning

Task-free online continual learning aims to alleviate catastrophic forgetting of the learner on a non-iid data stream. Experience Replay (ER) is a SOTA continual learning method, which is broadly used as the backbone algorithm for other replay-based methods. However, the training strategy of ER is too simple to take full advantage of replayed examples and its reservoir sampling strategy is also suboptimal. In this work, we propose a general proximal gradient framework so that ER can be viewed as a special case. We further propose two improvements accordingly: Principal Gradient Direction (PGD) and Confidence Reservoir Sampling (CRS). In Principal Gradient Direction, we optimize a target gradient that not only represents the major contribution of past gradients, but also retains the new knowledge of the current gradient. We then present Confidence Reservoir Sampling for maintaining a more informative memory buffer based on a margin-based metric that measures the value of stored examples. Experiments substantiate the effectiveness of both our improvements and our new algorithm consistently boosts the performance of MIR-replay, a SOTA ER-based method: our algorithm increases the average accuracy up to 7.9% and reduces forgetting up to 15.4% on four datasets.

cs.LG↗

arXiv2020

Contrastive Rendering for Ultrasound Image Segmentation

Ultrasound (US) image segmentation embraced its significant improvement in deep learning era. However, the lack of sharp boundaries in US images still remains an inherent challenge for segmentation. Previous methods often resort to global context, multi-scale cues or auxiliary guidance to estimate the boundaries. It is hard for these methods to approach pixel-level learning for fine-grained boundary generating. In this paper, we propose a novel and effective framework to improve boundary estimation in US images. Our work has three highlights. First, we propose to formulate the boundary estimation as a rendering task, which can recognize ambiguous points (pixels/voxels) and calibrate the boundary prediction via enriched feature representation learning. Second, we introduce point-wise contrastive learning to enhance the similarity of points from the same class and contrastively decrease the similarity of points from different classes. Boundary ambiguities are therefore further addressed. Third, both rendering and contrastive learning tasks contribute to consistent improvement while reducing network parameters. As a proof-of-concept, we performed validation experiments on a challenging dataset of 86 ovarian US volumes. Results show that our proposed method outperforms state-of-the-art methods and has the potential to be used in clinical practice.

eess.IV↗

arXiv2020

Synthesis and Edition of Ultrasound Images via Sketch Guided Progressive Growing GANs

Ultrasound (US) is widely accepted in clinic for anatomical structure inspection. However, lacking in resources to practice US scan, novices often struggle to learn the operation skills. Also, in the deep learning era, automated US image analysis is limited by the lack of annotated samples. Efficiently synthesizing realistic, editable and high resolution US images can solve the problems. The task is challenging and previous methods can only partially complete it. In this paper, we devise a new framework for US image synthesis. Particularly, we firstly adopt a sketch generative adversarial networks (Sgan) to introduce background sketch upon object mask in a conditioned generative adversarial network. With enriched sketch cues, Sgan can generate realistic US images with editable and fine-grained structure details. Although effective, Sgan is hard to generate high resolution US images. To achieve this, we further implant the Sgan into a progressive growing scheme (PGSgan). By smoothly growing both generator and discriminator, PGSgan can gradually synthesize US images from low to high resolution. By synthesizing ovary and follicle US images, our extensive perceptual evaluation, user study and segmentation results prove the promising efficacy and efficiency of the proposed PGSgan.

eess.IV↗

arXiv2020

Observation of high-temperature quantum anomalous Hall regime in intrinsic MnBi$_2$Te$_4$/Bi$_2$Te$_3$ superlattice

The quantum anomalous Hall effect is a fundamental transport response of a topologically non-trivial system in zero magnetic field. Its physical origin relies on the intrinsically inverted electronic band structure and ferromagnetism, and its most consequential manifestation is the dissipation-free flow of chiral charge currents at the edges that can potentially transform future quantum electronics. Here we report a previously unknown Berry-curvature-driven anomalous Hall regime ('Q-window') at above-Kelvin temperatures in the magnetic topological bulk crystals where through growth Mn ions self-organize into a period-ordered MnBi$_2$Te$_4$/Bi$_2$Te$_3$ superlattice. Robust ferromagnetism of the MnBi$_2$Te$_4$ monolayers opens a large surface gap, and anomalous Hall conductance reaches an $e^2/h$ quantization plateau when the Fermi level is tuned into this gap within a Q-window in which the anomalous Hall conductance from the bulk is to a high precision zero. The quantization in this new regime is not obstructed by the bulk conduction channels and thus should be present in a broad family of topological magnets.

cond-mat.mtrl-sci↗

arXiv2019

Spin memory of the topological material under strong disorder

Robustness to disorder - the defining property of any topological state - has been mostly tested in low-disorder translationally-invariant materials systems where the protecting underlying symmetry, such as time reversal, is preserved. The ultimate disorder limits to topological protection are still unknown, however, a number of theories predict that even in the amorphous state a quantized conductance might yet reemerge. Here we report a directly detected robust spin response in structurally disordered thin films of the topological material Sb2Te3 free of extrinsic magnetic dopants, which we controllably tune from a strong (amorphous) to a weak crystalline) disorder state. The magnetic signal onsets at a surprisingly high temperature (~ 200 K) and eventually ceases within the crystalline state. We demonstrate that in a strongly disordered state disorder-induced spin correlations dominate the transport of charge - they engender a spin memory phenomenon, generated by the nonequilibrium charge currents controlled by localized spins. The negative magnetoresistance (MR) in the extensive spin-memory phase space is isotropic. Within the crystalline state, it transitions into a positive MR corresponding to the weak antilocalization (WAL) quantum interference effect, with a 2D scaling characteristic of the topological state. Our findings demonstrate that these nonequilibrium currents set a disorder threshold to the topological state; they lay out a path to tunable spin-dependent charge transport and point to new possibilities of spin control by disorder engineering of topological materials

cond-mat.mtrl-sci↗