arXiv subjects

Wenda Zhou

Publications and source records attributed to Wenda Zhou.

16 recordsLinked to original sources

arXiv2025

Competitive Programming with Large Reasoning Models

We show that reinforcement learning applied to large language models (LLMs) significantly boosts performance on complex coding and reasoning tasks. Additionally, we compare two general-purpose reasoning models - OpenAI o1 and an early checkpoint of o3 - with a domain-specific system, o1-ioi, which uses hand-engineered inference strategies designed for competing in the 2024 International Olympiad in Informatics (IOI). We competed live at IOI 2024 with o1-ioi and, using hand-crafted test-time strategies, placed in the 49th percentile. Under relaxed competition constraints, o1-ioi achieved a gold medal. However, when evaluating later models such as o3, we find that o3 achieves gold without hand-crafted domain-specific strategies or relaxed constraints. Our findings show that although specialized pipelines such as o1-ioi yield solid improvements, the scaled-up, general-purpose o3 model surpasses those results without relying on hand-crafted inference heuristics. Notably, o3 achieves a gold medal at the 2024 IOI and obtains a Codeforces rating on par with elite human competitors. Overall, these results indicate that scaling general-purpose reinforcement learning, rather than relying on domain-specific techniques, offers a robust path toward state-of-the-art AI in reasoning domains, such as competitive programming.

cs.LG↗

arXiv2024

Versatile Cataract Fundus Image Restoration Model Utilizing Unpaired Cataract and High-quality Images

Cataract is one of the most common blinding eye diseases and can be treated by surgery. However, because cataract patients may also suffer from other blinding eye diseases, ophthalmologists must diagnose them before surgery. The cloudy lens of cataract patients forms a hazy degeneration in the fundus images, making it challenging to observe the patient's fundus vessels, which brings difficulties to the diagnosis process. To address this issue, this paper establishes a new cataract image restoration method named Catintell. It contains a cataract image synthesizing model, Catintell-Syn, and a restoration model, Catintell-Res. Catintell-Syn uses GAN architecture with fully unsupervised data to generate paired cataract-like images with realistic style and texture rather than the conventional Gaussian degradation algorithm. Meanwhile, Catintell-Res is an image restoration network that can improve the quality of real cataract fundus images using the knowledge learned from synthetic cataract images. Extensive experiments show that Catintell-Res outperforms other cataract image restoration methods in PSNR with 39.03 and SSIM with 0.9476. Furthermore, the universal restoration ability that Catintell-Res gained from unpaired cataract images can process cataract images from various datasets. We hope the models can help ophthalmologists identify other blinding eye diseases of cataract patients and inspire more medical image restoration methods in the future.

eess.IV↗

arXiv2022

pmwd: A Differentiable Cosmological Particle-Mesh $N$-body Library

The formation of the large-scale structure, the evolution and distribution of galaxies, quasars, and dark matter on cosmological scales, requires numerical simulations. Differentiable simulations provide gradients of the cosmological parameters, that can accelerate the extraction of physical information from statistical analyses of observational data. The deep learning revolution has brought not only myriad powerful neural networks, but also breakthroughs including automatic differentiation (AD) tools and computational accelerators like GPUs, facilitating forward modeling of the Universe with differentiable simulations. Because AD needs to save the whole forward evolution history to backpropagate gradients, current differentiable cosmological simulations are limited by memory. Using the adjoint method, with reverse time integration to reconstruct the evolution history, we develop a differentiable cosmological particle-mesh (PM) simulation library pmwd (particle-mesh with derivatives) with a low memory cost. Based on the powerful AD library JAX, pmwd is fully differentiable, and is highly performant on GPUs.

astro-ph.IM↗

arXiv2022

Vitruvion: A Generative Model of Parametric CAD Sketches

Parametric computer-aided design (CAD) tools are the predominant way that engineers specify physical structures, from bicycle pedals to airplanes to printed circuit boards. The key characteristic of parametric CAD is that design intent is encoded not only via geometric primitives, but also by parameterized constraints between the elements. This relational specification can be viewed as the construction of a constraint program, allowing edits to coherently propagate to other parts of the design. Machine learning offers the intriguing possibility of accelerating the design process via generative modeling of these structures, enabling new tools such as autocompletion, constraint inference, and conditional synthesis. In this work, we present such an approach to generative modeling of parametric CAD sketches, which constitute the basic computational building blocks of modern mechanical design. Our model, trained on real-world designs from the SketchGraphs dataset, autoregressively synthesizes sketches as sequences of primitives, with initial coordinates, and constraints that reference back to the sampled primitives. As samples from the model match the constraint graph representation used in standard CAD software, they may be directly imported, solved, and edited according to downstream design tasks. In addition, we condition the model on various contexts, including partial sketches (primers) and images of hand-drawn sketches. Evaluation of the proposed approach demonstrates its ability to synthesize realistic CAD sketches and its potential to aid the mechanical design workflow.

cs.LG↗

arXiv2022

Autobahn: Automorphism-based Graph Neural Nets

We introduce Automorphism-based graph neural networks (Autobahn), a new family of graph neural networks. In an Autobahn, we decompose the graph into a collection of subgraphs and apply local convolutions that are equivariant to each subgraph's automorphism group. Specific choices of local neighborhoods and subgraphs recover existing architectures such as message passing neural networks. Our formalism also encompasses novel architectures: as an example, we introduce a graph neural network that decomposes the graph into paths and cycles. The resulting convolutions reflect the natural way that parts of the graph can transform, preserving the intuitive meaning of convolution without sacrificing global permutation equivariance. We validate our approach by applying Autobahn to molecular graphs, where it achieves results competitive with state-of-the-art message passing algorithms.

cs.LG↗

arXiv2021

Compressed sensing in the presence of speckle noise

The problem of recovering a structured signal from its linear measurements in the presence of speckle noise is studied. This problem appears in many imaging systems such as synthetic aperture radar and optical coherence tomography. The current acquisition technology oversamples signals and converts the problem into a denoising problem with multiplicative noise. However, this paper explores the possibility of reducing the number of measurements below the ambient dimension of the signal. The sophistications that appear in the study of multiplicative noises have so far impeded theoretical analysis of such problems. This paper aims to present the first theoretical result regarding the recovery of signals from their undersampled measurements under the speckle noise. It is shown that if the signal class is structured, in the sense that the signals can be compressed efficiently, then one can obtain accurate estimates of the signal from fewer measurements than the ambient dimension. We demonstrate the effectiveness of the methods we propose through simulation results.

cs.IT↗

arXiv2020

Prediction for structure stability and ultrahigh hydrogen evolution performance of monolayer 2H-CrS2

By a combination of the first-principles calculations and climbing image nudged elastic band method (ciNEB) we investigate structure stabilities and hydrogen evolution reaction (HER) performance of monolayer 2H-CrS2. The results suggest the free energy for the Volmer reaction in the monolayer 2H-CrS2 with S vacancy is 0.07 eV, comparable with Pt-based catalyst, and HER on the surface of the monolayer is prone to the Volmer-Heyrovsky mechanism with no energy barrier. We propose that high HER performance stems from the reduction of the energy level of d-band center. Additionally, the S vacancy leads to defect states in the middle of electronic bandgap and the reduction of potential barrier between the S atom layer and the vacuum, which is conducive to improve HER performance.

cond-mat.mtrl-sci↗

arXiv2020

SketchGraphs: A Large-Scale Dataset for Modeling Relational Geometry in Computer-Aided Design

Parametric computer-aided design (CAD) is the dominant paradigm in mechanical engineering for physical design. Distinguished by relational geometry, parametric CAD models begin as two-dimensional sketches consisting of geometric primitives (e.g., line segments, arcs) and explicit constraints between them (e.g., coincidence, perpendicularity) that form the basis for three-dimensional construction operations. Training machine learning models to reason about and synthesize parametric CAD designs has the potential to reduce design time and enable new design workflows. Additionally, parametric CAD designs can be viewed as instances of constraint programming and they offer a well-scoped test bed for exploring ideas in program synthesis and induction. To facilitate this research, we introduce SketchGraphs, a collection of 15 million sketches extracted from real-world CAD models coupled with an open-source data processing pipeline. Each sketch is represented as a geometric constraint graph where edges denote designer-imposed geometric relationships between primitives, the nodes of the graph. We demonstrate and establish benchmarks for two use cases of the dataset: generative modeling of sketches and conditional generation of likely constraints given unconstrained geometry.

cs.LG↗

arXiv2020

Asymptotics of Cross-Validation

Cross validation is a central tool in evaluating the performance of machine learning and statistical models. However, despite its ubiquitous role, its theoretical properties are still not well understood. We study the asymptotic properties of the cross validated-risk for a large class of models. Under stability conditions, we establish a central limit theorem and Berry-Esseen bounds, which enable us to compute asymptotically accurate confidence intervals. Using our results, we paint a big picture for the statistical speed-up of cross validation compared to a train-test split procedure. A corollary of our results is that parametric M-estimators (or empirical risk minimizers) benefit from the "full" speed-up when performing cross-validation under the training loss. In other common cases, such as when the training is done using a surrogate loss or a regularizer, we show that the behavior of the cross-validated risk is complex with a variance reduction which may be smaller or larger than the "full" speed-up, depending on the model and the underlying distribution. We allow the number of folds to grow with the number of observations at any rate.

math.ST↗

arXiv2020

Error bounds in estimating the out-of-sample prediction error using leave-one-out cross validation in high-dimensions

We study the problem of out-of-sample risk estimation in the high dimensional regime where both the sample size $n$ and number of features $p$ are large, and $n/p$ can be less than one. Extensive empirical evidence confirms the accuracy of leave-one-out cross validation (LO) for out-of-sample risk estimation. Yet, a unifying theoretical evaluation of the accuracy of LO in high-dimensional problems has remained an open problem. This paper aims to fill this gap for penalized regression in the generalized linear family. With minor assumptions about the data generating process, and without any sparsity assumptions on the regression coefficients, our theoretical analysis obtains finite sample upper bounds on the expected squared error of LO in estimating the out-of-sample error. Our bounds show that the error goes to zero as $n,p \rightarrow \infty$, even when the dimension $p$ of the feature vectors is comparable with or greater than the sample size $n$. One technical advantage of the theory is that it can be used to clarify and connect some results from the recent literature on scalable approximate LO.

stat.ML↗

arXiv2019

Discrete Object Generation with Reversible Inductive Construction

The success of generative modeling in continuous domains has led to a surge of interest in generating discrete data such as molecules, source code, and graphs. However, construction histories for these discrete objects are typically not unique and so generative models must reason about intractably large spaces in order to learn. Additionally, structured discrete domains are often characterized by strict constraints on what constitutes a valid object and generative models must respect these requirements in order to produce useful novel samples. Here, we present a generative model for discrete objects employing a Markov chain where transitions are restricted to a set of local operations that preserve validity. Building off of generative interpretations of denoising autoencoders, the Markov chain alternates between producing 1) a sequence of corrupted objects that are valid but not from the data distribution, and 2) a learned reconstruction distribution that attempts to fix the corruptions while also preserving validity. This approach constrains the generative model to only produce valid objects, requires the learner to only discover local modifications to the objects, and avoids marginalization over an unknown and potentially large space of construction histories. We evaluate the proposed approach on two highly structured discrete domains, molecules and Laman graphs, and find that it compares favorably to alternative methods at capturing distributional statistics for a host of semantically relevant metrics.

cs.LG↗

arXiv2019

Non-Vacuous Generalization Bounds at the ImageNet Scale: A PAC-Bayesian Compression Approach

Modern neural networks are highly overparameterized, with capacity to substantially overfit to training data. Nevertheless, these networks often generalize well in practice. It has also been observed that trained networks can often be "compressed" to much smaller representations. The purpose of this paper is to connect these two empirical observations. Our main technical result is a generalization bound for compressed networks based on the compressed size. Combined with off-the-shelf compression algorithms, the bound leads to state of the art generalization guarantees; in particular, we provide the first non-vacuous generalization guarantees for realistic architectures applied to the ImageNet classification problem. As additional evidence connecting compression and generalization, we show that compressibility of models that tend to overfit is limited: We establish an absolute limit on expected compressibility as a function of expected generalization error, where the expectations are over the random choice of training examples. The bounds are complemented by empirical results that show an increase in overfitting implies an increase in the number of bits required to describe a trained network.

stat.ML↗

arXiv2019

Empirical Risk Minimization and Stochastic Gradient Descent for Relational Data

Empirical risk minimization is the main tool for prediction problems, but its extension to relational data remains unsolved. We solve this problem using recent ideas from graph sampling theory to (i) define an empirical risk for relational data and (ii) obtain stochastic gradients for this empirical risk that are automatically unbiased. This is achieved by considering the method by which data is sampled from a graph as an explicit component of model design. By integrating fast implementations of graph sampling schemes with standard automatic differentiation tools, we provide an efficient turnkey solver for the risk minimization problem. We establish basic theoretical properties of the procedure. Finally, we demonstrate relational ERM with application to two non-standard problems: one-stage training for semi-supervised node classification, and learning embedding vectors for vertex attributes. Experiments confirm that the turnkey inference procedure is effective in practice, and that the sampling scheme used for model specification has a strong effect on model performance. Code is available at https://github.com/wooden-spoon/relational-ERM.

stat.ML↗

arXiv2019

Denoising of structured random processes

Denoising stationary process $(X_i)_{i \in Z}$ corrupted by additive white Gaussian noise is a classic and fundamental problem in information theory and statistical signal processing. Despite considerable progress in designing efficient denoising algorithms, for general analog sources, theoretically-founded computationally-efficient methods are yet to be found. For instance in denoising $X^n$ corrupted by noise $Z^n$ as $Y^n=X^n+Z^n$, given the full distribution of $X^n$, a minimum mean square error (MMSE) denoiser needs to compute $E[X^n|Y^n]$. However, for general sources, computing $E[X^n|Y^n]$ is computationally very challenging, if not infeasible. In this paper, starting by a Bayesian setup, where the source distribution is fully known, a novel denoising method, namely, quantized maximum a posteriori (Q-MAP) denoiser, is proposed and its asymptotic performance in the high signal to noise ratio regime is analyzed. Both for memoryless sources, and for structured first-order Markov sources, it is shown that, asymptotically, as $σ$ converges to zero, ${1\over σ^2}E[(X_i-\hat{X}^{\rm Q-MAP}_i)^2]$ achieved by Q-MAP denoiser converges to the information dimension of the source. For the studied memoryless sources, this limit is known to be optimal. A key advantage of the Q-MAP denoiser is that, unlike an MMSE denoiser, it highlights the key properties of the source distribution that are to be used in its denoising. This property dramatically reduces the computational complexity of approximating the solution of the Q-MAP denoiser. Additionally, it naturally leads to a learning-based denoiser. Using ImageNet database for training, initial simulation results exploring the performance of such a learning-based denoiser in image denoising are presented.

cs.IT↗

arXiv2018

Approximate Leave-One-Out for High-Dimensional Non-Differentiable Learning Problems

Consider the following class of learning schemes: \begin{equation} \label{eq:main-problem1} \hat{\boldsymbolβ} := \underset{\boldsymbolβ \in \mathcal{C}}{\arg\min} \;\sum_{j=1}^n \ell(\boldsymbol{x}_j^\top\boldsymbolβ; y_j) + λR(\boldsymbolβ), \qquad \qquad \qquad (1) \end{equation} where $\boldsymbol{x}_i \in \mathbb{R}^p$ and $y_i \in \mathbb{R}$ denote the $i^{\rm th}$ feature and response variable respectively. Let $\ell$ and $R$ be the convex loss function and regularizer, $\boldsymbolβ$ denote the unknown weights, and $λ$ be a regularization parameter. $\mathcal{C} \subset \mathbb{R}^{p}$ is a closed convex set. Finding the optimal choice of $λ$ is a challenging problem in high-dimensional regimes where both $n$ and $p$ are large. We propose three frameworks to obtain a computationally efficient approximation of the leave-one-out cross validation (LOOCV) risk for nonsmooth losses and regularizers. Our three frameworks are based on the primal, dual, and proximal formulations of (1). Each framework shows its strength in certain types of problems. We prove the equivalence of the three approaches under smoothness conditions. This equivalence enables us to justify the accuracy of the three methods under such conditions. We use our approaches to obtain a risk estimate for several standard problems, including generalized LASSO, nuclear norm regularization, and support vector machines. We empirically demonstrate the effectiveness of our results for non-differentiable cases.

cs.LG↗

arXiv2018

Approximate Leave-One-Out for Fast Parameter Tuning in High Dimensions

Consider the following class of learning schemes: $$\hat{\boldsymbolβ} := \arg\min_{\boldsymbolβ}\;\sum_{j=1}^n \ell(\boldsymbol{x}_j^\top\boldsymbolβ; y_j) + λR(\boldsymbolβ),\qquad\qquad (1) $$ where $\boldsymbol{x}_i \in \mathbb{R}^p$ and $y_i \in \mathbb{R}$ denote the $i^{\text{th}}$ feature and response variable respectively. Let $\ell$ and $R$ be the loss function and regularizer, $\boldsymbolβ$ denote the unknown weights, and $λ$ be a regularization parameter. Finding the optimal choice of $λ$ is a challenging problem in high-dimensional regimes where both $n$ and $p$ are large. We propose two frameworks to obtain a computationally efficient approximation ALO of the leave-one-out cross validation (LOOCV) risk for nonsmooth losses and regularizers. Our two frameworks are based on the primal and dual formulations of (1). We prove the equivalence of the two approaches under smoothness conditions. This equivalence enables us to justify the accuracy of both methods under such conditions. We use our approaches to obtain a risk estimate for several standard problems, including generalized LASSO, nuclear norm regularization, and support vector machines. We empirically demonstrate the effectiveness of our results for non-differentiable cases.

stat.ML↗