arXiv subjects

Shuo Pang

Publications and source records attributed to Shuo Pang.

17 recordsLinked to original sources

arXiv2025

Lower Bounds for CSP Hierarchies Through Ideal Reduction

We present a generic way to obtain level lower bounds for (promise) CSP hierarchies from degree lower bounds for algebraic proof systems. More specifically, we show that pseudo-reduction operators in the sense of Alekhnovich and Razborov [Proc. Steklov Inst. Math. 2003] can be used to fool the cohomological $k$-consistency algorithm. As applications, we prove optimal level lower bounds for $c$ vs. $\ell$-coloring for all $\ell \geq c \geq 3$, and give a simplified proof of the lower bounds for lax and null-constraining CSPs of Chan and Ng [STOC 2025].

cs.CC↗

arXiv2025

Graph Colouring Is Hard on Average for Polynomial Calculus and Nullstellensatz

We prove that polynomial calculus (and hence also Nullstellensatz) over any field requires linear degree to refute that sparse random regular graphs, as well as sparse Erd\H{o}s-R\'{e}nyi random graphs, are $3$-colourable. Using the known relation between size and degree for polynomial calculus proofs, this implies strongly exponential lower bounds on proof size.

cs.CC↗

arXiv2024

Truly Supercritical Trade-offs for Resolution, Cutting Planes, Monotone Circuits, and Weisfeiler-Leman

We exhibit supercritical trade-off for monotone circuits, showing that there are functions computable by small circuits for which any circuit must have depth super-linear or even super-polynomial in the number of variables, far exceeding the linear worst-case upper bound. We obtain similar trade-offs in proof complexity, where we establish the first size-depth trade-offs for cutting planes and resolution that are truly supercritical, i.e., in terms of formula size rather than number of variables, and we also show supercritical trade-offs between width and size for treelike resolution. Our results build on a new supercritical width-depth trade-off for resolution, obtained by refining and strengthening the compression scheme for the Cop-Robber game in [Grohe, Lichter, Neuen & Schweitzer 2023]. This yields robust supercritical trade-offs for dimension versus iteration number in the Weisfeiler-Leman algorithm, which also translate into trade-offs between number of variables and quantifier depth in first-order logic. Our other results follow from improved lifting theorems that might be of independent interest.

cs.CC↗

arXiv2024

Sum-of-squares lower bounds for Non-Gaussian Component Analysis

Non-Gaussian Component Analysis (NGCA) is the statistical task of finding a non-Gaussian direction in a high-dimensional dataset. Specifically, given i.i.d.\ samples from a distribution $P^A_{v}$ on $\mathbb{R}^n$ that behaves like a known distribution $A$ in a hidden direction $v$ and like a standard Gaussian in the orthogonal complement, the goal is to approximate the hidden direction. The standard formulation posits that the first $k-1$ moments of $A$ match those of the standard Gaussian and the $k$-th moment differs. Under mild assumptions, this problem has sample complexity $O(n)$. On the other hand, all known efficient algorithms require $\Omega(n^{k/2})$ samples. Prior work developed sharp Statistical Query and low-degree testing lower bounds suggesting an information-computation tradeoff for this problem. Here we study the complexity of NGCA in the Sum-of-Squares (SoS) framework. Our main contribution is the first super-constant degree SoS lower bound for NGCA. Specifically, we show that if the non-Gaussian distribution $A$ matches the first $(k-1)$ moments of $\mathcal{N}(0, 1)$ and satisfies other mild conditions, then with fewer than $n^{(1 - \varepsilon)k/2}$ many samples from the normal distribution, with high probability, degree $(\log n)^{{1\over 2}-o_n(1)}$ SoS fails to refute the existence of such a direction $v$. Our result significantly strengthens prior work by establishing a super-polynomial information-computation tradeoff against a broader family of algorithms. As corollaries, we obtain SoS lower bounds for several problems in robust statistics and the learning of mixture models. Our SoS lower bound proof introduces a novel technique, that we believe may be of broader interest, and a number of refinements over existing methods.

cs.LG↗

arXiv2024

A mode-multiplexed photonic integrated vector dot-product core from inverse design

Photonic computing has the potential of harnessing the full degrees of freedom (DOFs) of the light field, including wavelength, spatial mode, spatial location, phase quadrature, and polarization, to achieve higher level of computation parallelization and scalability than digital electronic processors. While multiplexing using wavelength and other DOFs can be readily integrated on silicon photonics platforms with compact footprints, conventional mode-division multiplexed (MDM) photonic designs occupy areas exceeding tens to hundreds of microns for a few spatial modes, significantly limiting their scalability. Here we utilize inverse design to demonstrate an ultracompact photonic computing core that calculates vector dot-products based on MDM coherent mixing within a nominal footprint of 5 um x 3 um. Our dot-product core integrates the functionalities of 2 mode multiplexers and 1 multi-mode coherent mixers, all within the footprint, and could be applied to various computation and computer vision tasks, with high computing throughput density. We experimentally demonstrate computing examples on the fabricated core, including complex number multiplication and motion estimation using optical flow.

physics.optics↗

arXiv2023

Integrated Photonic Encoder for Terapixel Image Processing

Modern lens designs are capable of resolving >10 gigapixels, while advances in camera frame-rate and hyperspectral imaging have made Terapixel/s data acquisition a real possibility. The main bottlenecks preventing such high data-rate systems are power consumption and data storage. In this work, we show that analog photonic encoders could address this challenge, enabling high-speed image compression using orders-of-magnitude lower power than digital electronics. Our approach relies on a silicon-photonics front-end to compress raw image data, foregoing energy-intensive image conditioning and reducing data storage requirements. The compression scheme uses a passive disordered photonic structure to perform kernel-type random projections of the raw image data with minimal power consumption and low latency. A back-end neural network can then reconstruct the original images with structural similarity exceeding 90%. This scheme has the potential to process Terapixel/s data streams using less than 100 fJ/pixel, providing a path to ultra-high-resolution data and image acquisition systems.

physics.optics↗

arXiv2023

A physical neural network training approach toward multi-plane light conversion design

Multi-plane light converter (MPLC) designs supporting hundreds of modes are attractive in high-throughput optical communications. These photonic structures typically comprise >10 phase masks in free space, with millions of independent design parameters. Conventional MPLC design using wavefront matching updates one mask at a time while fixing the rest. Here we construct a physical neural network (PNN) to model the light propagation and phase modulation in MPLC, providing access to the entire parameter set for optimization, including not only profiles of the phase masks and the distances between them. PNN training supports flexible optimization sequences and is a superset of existing MPLC design methods. In addition, our method allows tuning of hyperparameters of PNN training such as learning rate and batch size. Because PNN-based MPLC is found to be insensitive to the number of input and target modes in each training step, we have demonstrated a high-order MPLC design (45 modes) using mini batches that fit into the available computing resources.

eess.SP↗

arXiv2022

A statistical reconstruction algorithm for positronium lifetime imaging using time-of-flight positron emission tomography

Positron emission tomography (PET) is an important modality for diagnosing diseases such as cancer and Alzheimer's disease, capable of revealing the uptake of radiolabeled molecules that target specific pathological markers of the diseases. Recently, positronium lifetime imaging (PLI) that adds to traditional PET the ability to explore properties of the tissue microenvironment beyond tracer uptake has been demonstrated with time-of-flight (TOF) PET and the use of non-pure positron emitters. However, achieving accurate reconstruction of lifetime images from data acquired by systems having a finite TOF resolution still presents a challenge. This paper focuses on the two-dimensional PLI, introducing a maximum likelihood estimation (MLE) method that employs an exponentially modified Gaussian (EMG) probability distribution that describes the positronium lifetime data produced by TOF PET. We evaluate the performance of our EMG-based MLE method against \st{traditional} approaches using exponential likelihood functions and penalized surrogate methods. Results from computer-simulated data reveal that the proposed EMG-MLE method can yield quantitatively accurate lifetime images. We also demonstrate that the proposed MLE formulation can be extended to handle PLI data containing multiple positron populations.

physics.med-ph↗

arXiv2021

Fixed-point iterative linear inverse solver with extended precision

Solving linear systems is a ubiquitous task in science and engineering. Because directly inverting a large-scale linear system can be computationally expensive, iterative algorithms are often used to numerically find the inverse. To accommodate the dynamic range and precision requirements, these iterative algorithms are often carried out on floating-point processing units. Low-precision, fixed-point processors require only a fraction of the energy per operation consumed by their floating-point counterparts, yet their current usages exclude iterative solvers due to the computational errors arising from fixed-point arithmetic. In this work, we show that for a simple iterative algorithm, such as Richardson iteration, using a fixed-point processor can provide the same rate of convergence and achieve high-precision solutions beyond its native precision limit when combined with residual iteration. These results indicate that power-efficient computing platform consisting of analog computing devices can be used to solve a broad range of problems without compromising the speed or precision.

math.NA↗

arXiv2020

Training of mixed-signal optical convolutional neural network with reduced quantization level

Mixed-signal artificial neural networks (ANNs) that employ analog matrix-multiplication accelerators can achieve higher speed and improved power efficiency. Though analog computing is known to be susceptible to noise and device imperfections, various analog computing paradigms have been considered as promising solutions to address the growing computing demand in machine learning applications, thanks to the robustness of ANNs. This robustness has been explored in low-precision, fixed-point ANN models, which have proven successful on compressing ANN model size on digital computers. However, these promising results and network training algorithms cannot be easily migrated to analog accelerators. The reason is that digital computers typically carry intermediate results with higher bit width, though the inputs and weights of each ANN layers are of low bit width; while the analog intermediate results have low precision, analogous to digital signals with a reduced quantization level. Here we report a training method for mixed-signal ANN with two types of errors in its analog signals, random noise, and deterministic errors (distortions). The results showed that mixed-signal ANNs trained with our proposed method can achieve an equivalent classification accuracy with noise level up to 50% of the ideal quantization step size. We have demonstrated this training method on a mixed-signal optical convolutional neural network based on diffractive optics.

eess.SP↗

arXiv2019

On CDCL-based proof systems with the ordered decision strategy

We prove that conflict-driven clause learning SAT-solvers with the ordered decision strategy and the DECISION learning scheme are equivalent to ordered resolution. We also prove that, by replacing this learning scheme with its opposite that stops after the first new clause when backtracking, they become equivalent to general resolution. To the best of our knowledge, this is the first theoretical study of the interplay between specific decision strategies and clause learning. For both results, we allow nondeterminism in the solver's ability to perform unit propagation, conflict analysis, and restarts, in a way that is similar to previous works in the literature. To aid the presentation of our results, and possibly future research, we define a model and language for discussing CDCL-based proof systems that allows for succinct and precise theorem statements.

cs.LO↗

arXiv2019

Signal retrieval with measurement system knowledge using variational generative model

Signal retrieval from a series of indirect measurements is a common task in many imaging, metrology and characterization platforms in science and engineering. Because most of the indirect measurement processes are well-described by physical models, signal retrieval can be solved with an iterative optimization that enforces measurement consistency and prior knowledge on the signal. These iterative processes are time-consuming and only accommodate a linear measurement process and convex signal constraints. Recently, neural networks have been widely adopted to supersede iterative signal retrieval methods by approximating the inverse mapping of the measurement model. However, networks with deterministic processes have failed to distinguish signal ambiguities in an ill-posed measurement system, and retrieved signals often lack consistency with the measurement. In this work we introduce a variational generative model to capture the distribution of all possible signals, given a particular measurement. By exploiting the known measurement model in the variational generative framework, our signal retrieval process resolves the ambiguity in the forward process, and learns to retrieve signals that satisfy the measurement with high fidelity in a variety of linear and nonlinear ill-posed systems, including ultrafast pulse retrieval, coded aperture compressive video sensing and image retrieval from Fresnel hologram.

eess.IV↗

arXiv2019

Photon allocation strategy in region-of-interest tomographic imaging

Photon counting detection is a promising approach toward effectively reducing the radiation dose in x-ray computed tomography (CT). Full CT reconstruction from a fraction of the detected photons required by scintillation-based detectors has been demonstrated. Current efforts in photon-counting CT have focused mainly on reconstruction techniques. In medical and industrial x-ray computed tomography (CT) applications, truncated projection from the region-of-interest (ROI) is another effective way of dose reduction, as information from the ROI is usually sufficient for diagnostic purpose. Projection truncation poses an ill-conditioned inverse problem, which can be improved by including projections from the exterior region. However, this trade-off between the interior reconstruction quality and the additional exterior measurement (extra dose) has not been studied. In this manuscript, we explore the number of detected x-ray photons as a new dimension for measurement engineering. Specifically, we design a flexible, photon-efficient measurement strategy for ROI reconstruction by incorporating the photon statistics at extremely low flux level (~10 photons per pixel). The optimized photon-allocation strategy shows 10 ~ 15-fold lower ROI reconstruction error than truncated projections, and 2-fold lower than whole-volume CT scan. Our analysis in few-photon interior tomography could serve as a new framework for dose-efficient, task-specific x-ray image acquisition design.

physics.med-ph↗

arXiv2018

Deep Learning Cell Imaging through Anderson Localizing Optical Fibre

We demonstrate a deep-learning-based fibre imaging system which can transfer real-time artifact-free cell images through a meter-long Anderson localizing optical fibre. The cell samples are illuminated by an incoherent LED light source. A deep convolutional neural network is applied to the image reconstruction process. The network training uses data generated by a set-up with straight fibre at room temperature (~20 {\deg}C) but can be utilized directly for high fidelity reconstruction of cell images that are transported through fibre with a few degrees bend and/or fibre with segments heated up to 50 {\deg}C. In addition, cell images located several millimeters away from the bare fibre end can be transported and recovered successfully without the assistance of any distal optics. We further evidence that the trained neural network is able to reconstruct the images of cells which are never used in the training process and feature very different morphology.

physics.optics↗

arXiv2018

Interior X-ray diffraction tomography with low-resolution exterior information

X-ray diffraction tomography (XDT) resolves spatially-variant XRD profiles within macroscopic objects, and provides improved material contrast compared to the conventional transmission-based computed tomography (CT). However, due to the small diffraction cross-section, XDT suffers from long imaging acquisition time, which could take tens of hours for a full scan using a table-top X-ray tube. In medical and industrial imaging applications, oftentimes only the XRD measurement within a region-of-interest (ROI) is required, which, together with the demand to reduce imaging time and radiation dose to the sample, motivates the development of interior XDT systems that scan and reconstruct only an internal region within the sample. The interior problem does not have a unique solution, and a direct inversion on the truncated projection data often leads to large reconstruction errors in ROI. To reduce the truncation artifacts, conventional attenuation-based interior reconstruction problems rely on a known region or piecewise constant constraint within the ROI, which do not apply to all the samples in XDT. Here we propose a quasi-interior XDT scheme that incorporates a small fraction of projection information from the exterior region to assist interior reconstruction. The low-resolution exterior projection data obviates the requirement for prior knowledge on the object, and allows the ROI reconstruction to be performed with the fast, widely-used filtered back-projection algorithm for easy integration into real-time XDT imaging modules.

physics.med-ph↗

arXiv2018

Few-photon computed x-ray imaging

X-ray is a ubiquitous imaging modality in clinical diagnostics and industrial inspections, thanks to its high penetration power. Conventional x-ray imaging system, equipped with energy-integrating detectors, collects approximately 1000 to 10000 counts per pixel to ensure sufficient signal to noise ratio (SNR). The recent development of energy sensitive photon counting detectors opens new possibilities for x-ray imaging at low photon flux. In this letter, we report a novel photon-counting scheme that records the time stamp of individual photons, which follows a negative binomial distribution, and demonstrated the reconstruction based on the few-photon statistics. The projection and tomography reconstruction from measurements of roughly 10 photons shows the potential of using photon counting detectors for dose-efficient x-ray imaging systems.

physics.med-ph↗

arXiv2012

Large-field-of-view Chip-scale Talbot-grid-based Fluorescence Microscopy

The fluorescence microscope is one of the most important tools in modern clinical diagnosis and biological science. However, its expense, size and limited field-of-view (FOV) are becoming bottlenecks in key applications such as large-scale phenotyping and low-resource-setting diagnostics. Here we report a low-cost, compact chip-scale fluorescence-imaging platform, termed the Fluorescence Talbot Microscopy (FTM), which utilizes the Talbot self-imaging effect to enable efficient fluorescence imaging over a large and directly-scalable FOV. The FTM prototype has a resolution of 1.2 microns and an FOV of 3.9 mm x 3.5 mm. We demonstrate the imaging capability of FTM on fluorescently labeled breast cancer cells (SK-BR-3) and HEK cells expressing green fluorescent protein.

physics.optics↗