arXiv subjects

Yichun Zhou

Publications and source records attributed to Yichun Zhou.

8 recordsLinked to original sources

arXiv2026

SteerFace: Debiasing Synthetic Face Generation via Adaptive Residue Perturbation

The shortage of legally compliant data for face recognition training has sparked growing interest in using synthetic data as an alternative. While recent diffusion-based methods enable the generation of photorealistic face images with strong identity adherence and data diversity, their downstream recognition performance still exhibits a significant synthetic-real gap. This paper identifies visual tendency as a previously underexplored limitation, whereby synthetic data exhibit an unrealistic prevalence of visual attributes and thus deviate from the real-data distribution. Visual tendency can be attributed to the generator's conditioning on identity embeddings, through which co-occurring residual visual cues are unintentionally absorbed into learned identity semantics. To discourage the generator from exploiting such visual cues, this paper proposes SteerFace, a simple and efficient training framework that perturbs identity embeddings by steering them toward random orthogonal directions on the embedding hypersphere. The perturbation serves as an identity-preserving regularizer that penalizes the generator's reliance on non-identity components, as supported by theoretical analysis. This paper further introduces an adaptive strategy that learns perturbation strengths with both sample-wise preference and favorable overall statistics. Extensive experiments show that SteerFace effectively mitigates visual tendency, outperforms prior methods in downstream face recognition, and generalizes well across different training datasets and generation pipelines.

cs.CV↗

arXiv2026

Building an Affordable Self-Driving Lab: Practical Machine Learning Experiments for Physics Education Using Internet-of-Things

Machine learning (ML) is transforming modern physics research, but practical, hands-on experience with ML techniques remains limited due to cost and complexity barriers. To address this gap, we introduce an affordable, autonomous, Internet-of-Things (IoT)-enabled experimental platform designed specifically for applied physics education. Utilizing an Arduino microcontroller, a customizable multi-wavelength light emitting diode (LED) array, and photosensors, our setup generates diverse, real-time optical datasets ideal for training and evaluating foundational ML algorithms, including traversal methods, Bayesian inference, and deep learning. The platform facilitates a closed-loop, self-driving experimental workflow, encompassing automated data collection, preprocessing, model training, and validation. Through systematic performance comparisons, we demonstrate the superior ability of deep learning to capture complex nonlinear relationships compared to traversal and Bayesian methods. At approximately $60, this open-source IoT platform provides an accessible, practical pathway for students to master advanced ML concepts, promoting deeper conceptual insights and essential technical skills required for the next generation of physicists and engineers.

physics.ed-ph↗

arXiv2023

Unconventional polarization switching mechanism in (Hf, Zr)O2 ferroelectrics

HfO$_{2}$-based ferroelectric thin films are promising for their application in ferroelectric devices. Predicting the ultimate magnitude of polarization and understanding its switching mechanism are critical to realize the optimal performance of these devices. Here, a generalized solid-state variable cell nudged elastic band (VCNEB) method is employed to predict the switching pathway associated with domain-wall motion in (Hf, Zr)O$_{2}$ ferroelectrics. It is found that the polarization reversal pathway, where three-fold coordinated O atoms pass across the nominal unit-cell boundaries defined by the Hf/Zr atomic planes, is energetically more favorable than the conventional pathway where the O atoms do not pass through these planes. This finding implies that the polarization orientation in the orthorhombic Pca2$_{1}$ phase of HfO$_{2}$ nd its derivatives is opposite to that normally assumed, predicts the spontaneous polarization magnitude of about 70 $μ$C/cm$^{2}$ that is nearly 50% larger than the commonly accepted value, signifies a positive intrinsic longitudinal piezoelectric coefficient, and suggests growth of ferroelectric domains, in response to an applied electric field, structurally reversed to those usually anticipated. These results provide important insights into the understanding of ferroelectricity in HfO$_{2}$-based ferroelectrics.

cond-mat.mtrl-sci↗

arXiv2022

Bamboo: Building Mega-Scale Vision Dataset Continually with Human-Machine Synergy

Large-scale datasets play a vital role in computer vision. But current datasets are annotated blindly without differentiation to samples, making the data collection inefficient and unscalable. The open question is how to build a mega-scale dataset actively. Although advanced active learning algorithms might be the answer, we experimentally found that they are lame in the realistic annotation scenario where out-of-distribution data is extensive. This work thus proposes a novel active learning framework for realistic dataset annotation. Equipped with this framework, we build a high-quality vision dataset -- Bamboo, which consists of 69M image classification annotations with 119K categories and 28M object bounding box annotations with 809 categories. We organize these categories by a hierarchical taxonomy integrated from several knowledge bases. The classification annotations are four times larger than ImageNet22K, and that of detection is three times larger than Object365. Compared to ImageNet22K and Objects365, models pre-trained on Bamboo achieve superior performance among various downstream tasks (6.2% gains on classification and 2.1% gains on detection). We believe our active learning framework and Bamboo are essential for future work.

cs.CV↗

arXiv2021

ForgeryNet: A Versatile Benchmark for Comprehensive Forgery Analysis

The rapid progress of photorealistic synthesis techniques has reached at a critical point where the boundary between real and manipulated images starts to blur. Thus, benchmarking and advancing digital forgery analysis have become a pressing issue. However, existing face forgery datasets either have limited diversity or only support coarse-grained analysis. To counter this emerging threat, we construct the ForgeryNet dataset, an extremely large face forgery dataset with unified annotations in image- and video-level data across four tasks: 1) Image Forgery Classification, including two-way (real / fake), three-way (real / fake with identity-replaced forgery approaches / fake with identity-remained forgery approaches), and n-way (real and 15 respective forgery approaches) classification. 2) Spatial Forgery Localization, which segments the manipulated area of fake images compared to their corresponding source real images. 3) Video Forgery Classification, which re-defines the video-level forgery classification with manipulated frames in random positions. This task is important because attackers in real world are free to manipulate any target frame. and 4) Temporal Forgery Localization, to localize the temporal segments which are manipulated. ForgeryNet is by far the largest publicly available deep face forgery dataset in terms of data-scale (2.9 million images, 221,247 videos), manipulations (7 image-level approaches, 8 video-level approaches), perturbations (36 independent and more mixed perturbations) and annotations (6.3 million classification labels, 2.9 million manipulated area annotations and 221,247 temporal forgery segment labels). We perform extensive benchmarking and studies of existing face forensics methods and obtain several valuable observations.

cs.CV↗

arXiv2014

A Supersolid Skin Covering both Water and Ice

The mysterious nature and functionality of water and ice skins remain baffling to the community since 1859 when Farady firstly proposed liquid skin lubricating ice. Here we show the presence of supersolid phase that covers both water and ice using Raman spectroscopy measurements and quantum calculations. In the skin of two molecular layers thick, molecular undercoordination shortens the H-O bond by ~16% and lengthens the OH nonbond by ~25% through repulsion between electron pairs on adjacent O atoms, which depresses the density from 0.92 for bulk ice to 0.75 gcm-3. The O:H-O cooperative relaxation stiffens the H-O stretching phonon from 3200/3150 cm-1 to the same value of 3450 cm-1 and raises the melting temperature of both skins by up to ~310 K. Numerical derivatives on the viscosity and charge accumulation suggests that the elastic, polarized, and thermally stable supersolid phase makes the ice frictionless and water skin hydrophobic and ice like at room temperature.

cond-mat.soft↗

arXiv2013

Skin supersolidity slipperizing ice

Consistency between theory predictions and measurements and calculations revealed that the skin of ice, containing water molecules with fewer than four neighbours, forms a supersolid phase that is highly polarized, elastic, hydrophobic, with ultra-low density and high thermal stability. The supersolidity of skin sliperizes ice.

cond-mat.soft↗

arXiv2013

Mediation of hydrogen-bond coupling interactions by programmable heating and salting

We show that programmable heating and salting share the same effect on the frequency shift of the O:H and the H-O stretching phonons of the O:H-O hydrogen bond, which revealed that both heating and salting lengthens and softens the O:H bond and shortens and stiffens the H-O bond due to the weakening of the Coulomb repulsion between electron pairs of adjacent oxygen atoms. Understanding provides possible mechanism for the Hofmeister series and the detergent effect on cloth cleaning.

cond-mat.soft↗