arXiv subjects

Cheng Zhang

Publications and source records attributed to Cheng Zhang.

At least 307 records · Page 17Linked to original sources

arXiv2023

Hierarchical Semi-Implicit Variational Inference with Application to Diffusion Model Acceleration

Semi-implicit variational inference (SIVI) has been introduced to expand the analytical variational families by defining expressive semi-implicit distributions in a hierarchical manner. However, the single-layer architecture commonly used in current SIVI methods can be insufficient when the target posterior has complicated structures. In this paper, we propose hierarchical semi-implicit variational inference, called HSIVI, which generalizes SIVI to allow more expressive multi-layer construction of semi-implicit distributions. By introducing auxiliary distributions that interpolate between a simple base distribution and the target distribution, the conditional layers can be trained by progressively matching these auxiliary distributions one layer after another. Moreover, given pre-trained score networks, HSIVI can be used to accelerate the sampling process of diffusion models with the score matching objective. We show that HSIVI significantly enhances the expressiveness of SIVI on several Bayesian inference problems with complicated target distributions. When used for diffusion model acceleration, we show that HSIVI can produce high quality samples comparable to or better than the existing fast diffusion model based samplers with a small number of function evaluations on various datasets.

cs.LG↗

arXiv2023

Particle-based Variational Inference with Generalized Wasserstein Gradient Flow

Particle-based variational inference methods (ParVIs) such as Stein variational gradient descent (SVGD) update the particles based on the kernelized Wasserstein gradient flow for the Kullback-Leibler (KL) divergence. However, the design of kernels is often non-trivial and can be restrictive for the flexibility of the method. Recent works show that functional gradient flow approximations with quadratic form regularization terms can improve performance. In this paper, we propose a ParVI framework, called generalized Wasserstein gradient descent (GWG), based on a generalized Wasserstein gradient flow of the KL divergence, which can be viewed as a functional gradient method with a broader class of regularizers induced by convex functions. We show that GWG exhibits strong convergence guarantees. We also provide an adaptive version that automatically chooses Wasserstein metric to accelerate convergence. In experiments, we demonstrate the effectiveness and efficiency of the proposed framework on both simulated and real data problems.

stat.ML↗

arXiv2023

Revisiting Block-based Quantisation: What is Important for Sub-8-bit LLM Inference?

The inference of Large language models (LLMs) requires immense computation and memory resources. To curtail these costs, quantisation has merged as a promising solution, but existing LLM quantisation mainly focuses on 8-bit. In this work, we explore the statistical and learning properties of the LLM layer and attribute the bottleneck of LLM quantisation to numerical scaling offsets. To address this, we adapt block quantisations for LLMs, a family of methods that share scaling factors across packed numbers. Block quantisations efficiently reduce the numerical scaling offsets solely from an arithmetic perspective, without additional treatments in the computational path. Our nearly-lossless quantised 6-bit LLMs achieve a $19\times$ higher arithmetic density and $5\times$ memory density than the float32 baseline, surpassing the prior art 8-bit quantisation by $2.5\times$ in arithmetic density and $1.2\times$ in memory density, without requiring any data calibration or re-training. We also share our insights into sub-8-bit LLM quantisation, including the mismatch between activation and weight distributions, optimal fine-tuning strategies, and a lower quantisation granularity inherent in the statistical properties of LLMs. The latter two tricks enable nearly-lossless 4-bit LLMs on downstream tasks. Our code is open-sourced.

cs.LG↗

arXiv2023

QCD resummation of dijet azimuthal decorrelations in pp and pA collisions

We study the azimuthal angular decorrelations of dijet production in both proton-proton (pp) and proton-nucleus (pA) collisions. By utilizing soft-collinear effective theory, we establish the factorization and resummation formalism at the next-to-leading logarithmic accuracy for the azimuthal angular decorrelations in the back-to-back limit in pp collisions. We propose an approach where the nuclear modifications to dijet production in pA collisions are accounted for in the nuclear modified transverse momentum dependent parton distribution functions (nTMDPDFs), which contain both collinear and transverse dynamics. This approach naturally generalizes the well-established formalism related to the nuclear modified collinear parton distribution functions (nPDFs). We demonstrate strong consistency between our methodology and the CMS measurements in both pp and pA collisions, and make predictions for dijet production in the forward rapidity region in pA collisions at LHC kinematics and for mid-rapidity kinematics at sPHENIX. Throughout this paper, we focus on the application of this formalism to a simultaneous fit to both collinear and transverse momentum dependent contributions to the transverse momentum dependent distributions.

hep-ph↗

arXiv2023

ARTree: A Deep Autoregressive Model for Phylogenetic Inference

Designing flexible probabilistic models over tree topologies is important for developing efficient phylogenetic inference methods. To do that, previous works often leverage the similarity of tree topologies via hand-engineered heuristic features which would require pre-sampled tree topologies and may suffer from limited approximation capability. In this paper, we propose a deep autoregressive model for phylogenetic inference based on graph neural networks (GNNs), called ARTree. By decomposing a tree topology into a sequence of leaf node addition operations and modeling the involved conditional distributions based on learnable topological features via GNNs, ARTree can provide a rich family of distributions over the entire tree topology space that have simple sampling algorithms and density estimation procedures, without using heuristic features. We demonstrate the effectiveness and efficiency of our method on a benchmark of challenging real data tree topology density estimation and variational Bayesian phylogenetic inference problems.

q-bio.PE↗

arXiv2023

Deep learning models for price forecasting of financial time series: A review of recent advancements: 2020-2022

Accurately predicting the prices of financial time series is essential and challenging for the financial sector. Owing to recent advancements in deep learning techniques, deep learning models are gradually replacing traditional statistical and machine learning models as the first choice for price forecasting tasks. This shift in model selection has led to a notable rise in research related to applying deep learning models to price forecasting, resulting in a rapid accumulation of new knowledge. Therefore, we conducted a literature review of relevant studies over the past three years with a view to aiding researchers and practitioners in the field. This review delves deeply into deep learning-based forecasting models, presenting information on model architectures, practical applications, and their respective advantages and disadvantages. In particular, detailed information is provided on advanced models for price forecasting, such as Transformers, generative adversarial networks (GANs), graph neural networks (GNNs), and deep quantum neural networks (DQNNs). The present contribution also includes potential directions for future research, such as examining the effectiveness of deep learning models with complex structures for price forecasting, extending from point prediction to interval prediction using deep learning models, scrutinising the reliability and validity of decomposition ensembles, and exploring the influence of data volume on model performance.

q-fin.ST↗

arXiv2023

Traffic-Aware Hierarchical Beam Selection for Cell-Free Massive MIMO

Beam selection for joint transmission in cell-free massive multi-input multi-output systems faces the problem of extremely high training overhead and computational complexity. The traffic-aware quality of service additionally complicates the beam selection problem. To address this issue, we propose a traffic-aware hierarchical beam selection scheme performed in a dual timescale. In the long-timescale, the central processing unit collects wide beam responses from base stations (BSs) to predict the power profile in the narrow beam space with a convolutional neural network, based on which the cascaded multiple-BS beam space is carefully pruned. In the short-timescale, we introduce a centralized reinforcement learning (RL) algorithm to maximize the satisfaction rate of delay w.r.t. beam selection within multiple consecutive time slots. Moreover, we put forward three scalable distributed algorithms including hierarchical distributed Lyapunov optimization, fully distributed RL, and centralized training with decentralized execution of RL to achieve better scalability and better tradeoff between the performance and the execution signal overhead. Numerical results demonstrate that the proposed schemes significantly reduce both model training cost and beam training overhead and are easier to meet the user-specific delay requirement, compared to existing methods.

cs.IT↗

arXiv2023

ITI-GEN: Inclusive Text-to-Image Generation

Text-to-image generative models often reflect the biases of the training data, leading to unequal representations of underrepresented groups. This study investigates inclusive text-to-image generative models that generate images based on human-written prompts and ensure the resulting images are uniformly distributed across attributes of interest. Unfortunately, directly expressing the desired attributes in the prompt often leads to sub-optimal results due to linguistic ambiguity or model misrepresentation. Hence, this paper proposes a drastically different approach that adheres to the maxim that "a picture is worth a thousand words". We show that, for some attributes, images can represent concepts more expressively than text. For instance, categories of skin tones are typically hard to specify by text but can be easily represented by example images. Building upon these insights, we propose a novel approach, ITI-GEN, that leverages readily available reference images for Inclusive Text-to-Image GENeration. The key idea is learning a set of prompt embeddings to generate images that can effectively represent all desired attribute categories. More importantly, ITI-GEN requires no model fine-tuning, making it computationally efficient to augment existing text-to-image models. Extensive experiments demonstrate that ITI-GEN largely improves over state-of-the-art models to generate inclusive images from a prompt. Project page: https://czhang0528.github.io/iti-gen.

cs.CV↗

arXiv2023

Semi-Implicit Variational Inference via Score Matching

Semi-implicit variational inference (SIVI) greatly enriches the expressiveness of variational families by considering implicit variational distributions defined in a hierarchical manner. However, due to the intractable densities of variational distributions, current SIVI approaches often use surrogate evidence lower bounds (ELBOs) or employ expensive inner-loop MCMC runs for unbiased ELBOs for training. In this paper, we propose SIVI-SM, a new method for SIVI based on an alternative training objective via score matching. Leveraging the hierarchical structure of semi-implicit variational families, the score matching objective allows a minimax formulation where the intractable variational densities can be naturally handled with denoising score matching. We show that SIVI-SM closely matches the accuracy of MCMC and outperforms ELBO-based SIVI methods in a variety of Bayesian inference tasks.

stat.ML↗

arXiv2023

Incremental Collaborative Beam Alignment for Millimeter Wave Cell-Free MIMO Systems

Millimeter wave (mmWave) cell-free MIMO achieves an extremely high rate while its beam alignment (BA) suffers from excessive overhead due to a large number of transceivers. Recently, user location and probing measurements are utilized for BA based on machine learning (ML) models, e.g., deep neural network (DNN). However, most of these ML models are centralized with high communication and computational overhead and give no specific consideration to practical issues, e.g., limited training data and real-time model updates. In this paper, we study the {probing} beam-based BA for mmWave cell-free MIMO downlink with the help of broad learning (BL). For channels without and with uplink-downlink reciprocity, we propose the user-side and base station (BS)-side BL-aided incremental collaborative BA approaches. Via transforming the centralized BL into a distributed learning with data and feature splitting respectively, the user-side and BS-side schemes realize implicit sharing of multiple user data and multiple BS features. Simulations confirm that the user-side scheme is applicable to fast time-varying and/or non-stationary channels, while the BS-side scheme is suitable for systems with low-bandwidth fronthaul links and a central unit with limited computing power. The advantages of proposed schemes are also demonstrated compared to traditional and DNN-aided BA schemes.

cs.IT↗

arXiv2023

All-in-one Multi-degradation Image Restoration Network via Hierarchical Degradation Representation

The aim of image restoration is to recover high-quality images from distorted ones. However, current methods usually focus on a single task (\emph{e.g.}, denoising, deblurring or super-resolution) which cannot address the needs of real-world multi-task processing, especially on mobile devices. Thus, developing an all-in-one method that can restore images from various unknown distortions is a significant challenge. Previous works have employed contrastive learning to learn the degradation representation from observed images, but this often leads to representation drift caused by deficient positive and negative pairs. To address this issue, we propose a novel All-in-one Multi-degradation Image Restoration Network (AMIRNet) that can effectively capture and utilize accurate degradation representation for image restoration. AMIRNet learns a degradation representation for unknown degraded images by progressively constructing a tree structure through clustering, without any prior knowledge of degradation information. This tree-structured representation explicitly reflects the consistency and discrepancy of various distortions, providing a specific clue for image restoration. To further enhance the performance of the image restoration network and overcome domain gaps caused by unknown distortions, we design a feature transform block (FTB) that aligns domains and refines features with the guidance of the degradation representation. We conduct extensive experiments on multiple distorted datasets, demonstrating the effectiveness of our method and its advantages over state-of-the-art restoration methods both qualitatively and quantitatively.

cs.CV↗

arXiv2023

Robust transitionless quantum driving: Concatenated approach

We propose a concatenated approach for implementing transitionless quantum driving regardless of adiabatic conditions while being robustness with respect to all kinds of systematic errors induced by pulse duration, pulse amplitude, detunings, and Stark shift, etc. The current approach is particularly efficient for all time-dependent pulses with arbitrary shape, and only the phase differences between pulses is required to properly modulate. The simple physical implementation without the help of pulse shaping techniques or extra pulses makes this approach quite universal and provides a different avenue for robust quantum control by the time-dependent Hamiltonian.

quant-ph↗

arXiv2023

CO-BED: Information-Theoretic Contextual Optimization via Bayesian Experimental Design

We formalize the problem of contextual optimization through the lens of Bayesian experimental design and propose CO-BED -- a general, model-agnostic framework for designing contextual experiments using information-theoretic principles. After formulating a suitable information-based objective, we employ black-box variational methods to simultaneously estimate it and optimize the designs in a single stochastic gradient scheme. In addition, to accommodate discrete actions within our framework, we propose leveraging continuous relaxation schemes, which can naturally be integrated into our variational objective. As a result, CO-BED provides a general and automated solution to a wide range of contextual optimization problems. We illustrate its effectiveness in a number of experiments, where CO-BED demonstrates competitive performance even when compared to bespoke, model-specific alternatives.

stat.ML↗

arXiv2023

Doping transition-metal atoms in graphene for atomic-scale tailoring of electronic, magnetic, and quantum topological properties

Atomic-scale fabrication is an outstanding challenge and overarching goal for the nanoscience community. The practical implementation of moving and fixing atoms to a structure is non-trivial considering that one must spatially address the positioning of single atoms, provide a stabilizing scaffold to hold structures in place, and understand the details of their chemical bonding. Free-standing graphene offers a simplified platform for the development of atomic-scale fabrication and the focused electron beam in a scanning transmission electron microscope can be used to locally induce defects and sculpt the graphene. In this scenario, the graphene forms the stabilizing scaffold and the experimental question is whether a range of dopant atoms can be attached and incorporated into the lattice using a single technique and, from a theoretical perspective, we would like to know which dopants will create technologically interesting properties. Here, we demonstrate that the electron beam can be used to selectively and precisely insert a variety of transition metal atoms into graphene with highly localized control over the doping locations. We use first-principles density functional theory calculations with direct observation of the created structures to reveal the energetics of incorporating metal atoms into graphene and their magnetic, electronic, and quantum topological properties.

cond-mat.mtrl-sci↗

arXiv2023

Cloud-Native Computing: A Survey from the Perspective of Services

The development of cloud computing delivery models inspires the emergence of cloud-native computing. Cloud-native computing, as the most influential development principle for web applications, has already attracted increasingly more attention in both industry and academia. Despite the momentum in the cloud-native industrial community, a clear research roadmap on this topic is still missing. As a contribution to this knowledge, this paper surveys key issues during the life-cycle of cloud-native applications, from the perspective of services. Specifically, we elaborate the research domains by decoupling the life-cycle of cloud-native applications into four states: building, orchestration, operate, and maintenance. We also discuss the fundamental necessities and summarize the key performance metrics that play critical roles during the development and management of cloud-native applications. We highlight the key implications and limitations of existing works in each state. The challenges, future directions, and research opportunities are also discussed.

cs.DC↗

arXiv2023

OVO: Open-Vocabulary Occupancy

Semantic occupancy prediction aims to infer dense geometry and semantics of surroundings for an autonomous agent to operate safely in the 3D environment. Existing occupancy prediction methods are almost entirely trained on human-annotated volumetric data. Although of high quality, the generation of such 3D annotations is laborious and costly, restricting them to a few specific object categories in the training dataset. To address this limitation, this paper proposes Open Vocabulary Occupancy (OVO), a novel approach that allows semantic occupancy prediction of arbitrary classes but without the need for 3D annotations during training. Keys to our approach are (1) knowledge distillation from a pre-trained 2D open-vocabulary segmentation model to the 3D occupancy network, and (2) pixel-voxel filtering for high-quality training data generation. The resulting framework is simple, compact, and compatible with most state-of-the-art semantic occupancy prediction models. On NYUv2 and SemanticKITTI datasets, OVO achieves competitive performance compared to supervised semantic occupancy prediction approaches. Furthermore, we conduct extensive analyses and ablation studies to offer insights into the design of the proposed framework. Our code is publicly available at https://github.com/dzcgaara/OVO.

cs.CV↗

arXiv2023

Lepton pair production in UPCs: towards the precision test of the resummation formalism

We present a detailed investigation of the azimuthal asymmetries and acoplanarity in lepton pair production in ultraperipheral collisions (UPCs). These observables provide a unique opportunity to test the SCET resummation formalism, given the extremely high photon flux in UPCs, which enables precise measurements of these processes. We improve the accuracy of the previous calculations by including the soft photon contributions beyond the double leading logarithm approximation. Notably, the single logarithm terms arising from the collinear region are greatly enhanced by the small mass of the leptons. Our findings demonstrate the accessibility of these sub-leading logarithm resummation effects through the analysis of angular correlations in lepton pairs produced in UPCs at the Relativistic Heavy Ion Collider (RHIC) and the Large Hadron Collider (LHC).

hep-ph↗

arXiv2023

One for All: Unified Workload Prediction for Dynamic Multi-tenant Edge Cloud Platforms

Workload prediction in multi-tenant edge cloud platforms (MT-ECP) is vital for efficient application deployment and resource provisioning. However, the heterogeneous application patterns, variable infrastructure performance, and frequent deployments in MT-ECP pose significant challenges for accurate and efficient workload prediction. Clustering-based methods for dynamic MT-ECP modeling often incur excessive costs due to the need to maintain numerous data clusters and models, which leads to excessive costs. Existing end-to-end time series prediction methods are challenging to provide consistent prediction performance in dynamic MT-ECP. In this paper, we propose an end-to-end framework with global pooling and static content awareness, DynEformer, to provide a unified workload prediction scheme for dynamic MT-ECP. Meticulously designed global pooling and information merging mechanisms can effectively identify and utilize global application patterns to drive local workload predictions. The integration of static content-aware mechanisms enhances model robustness in real-world scenarios. Through experiments on five real-world datasets, DynEformer achieved state-of-the-art in the dynamic scene of MT-ECP and provided a unified end-to-end prediction scheme for MT-ECP.

cs.NI↗