arXiv subjects

Pengcheng Wu

Publications and source records attributed to Pengcheng Wu.

36 records · Page 2Linked to original sources

arXiv2025

HYDRA: Hypergradient Data Relevance Analysis for Interpreting Deep Neural Networks

The behaviors of deep neural networks (DNNs) are notoriously resistant to human interpretations. In this paper, we propose Hypergradient Data Relevance Analysis, or HYDRA, which interprets the predictions made by DNNs as effects of their training data. Existing approaches generally estimate data contributions around the final model parameters and ignore how the training data shape the optimization trajectory. By unrolling the hypergradient of test loss w.r.t. the weights of training data, HYDRA assesses the contribution of training data toward test data points throughout the training trajectory. In order to accelerate computation, we remove the Hessian from the calculation and prove that, under moderate conditions, the approximation error is bounded. Corroborating this theoretical claim, empirical results indicate the error is indeed small. In addition, we quantitatively demonstrate that HYDRA outperforms influence functions in accurately estimating data contribution and detecting noisy data labels. The source code is available at https://github.com/cyyever/aaai_hydra_8686.

cs.LG↗

arXiv2025

FedOBD: Opportunistic Block Dropout for Efficiently Training Large-scale Neural Networks through Federated Learning

Large-scale neural networks possess considerable expressive power. They are well-suited for complex learning tasks in industrial applications. However, large-scale models pose significant challenges for training under the current Federated Learning (FL) paradigm. Existing approaches for efficient FL training often leverage model parameter dropout. However, manipulating individual model parameters is not only inefficient in meaningfully reducing the communication overhead when training large-scale FL models, but may also be detrimental to the scaling efforts and model performance as shown by recent research. To address these issues, we propose the Federated Opportunistic Block Dropout (FedOBD) approach. The key novelty is that it decomposes large-scale models into semantic blocks so that FL participants can opportunistically upload quantized blocks, which are deemed to be significant towards training the model, to the FL server for aggregation. Extensive experiments evaluating FedOBD against four state-of-the-art approaches based on multiple real-world datasets show that it reduces the overall communication overhead by more than 88% compared to the best performing baseline approach, while achieving the highest test accuracy. To the best of our knowledge, FedOBD is the first approach to perform dropout on FL models at the block level rather than at the individual parameter level.

cs.LG↗

arXiv2025

Self-Bootstrapping for Versatile Test-Time Adaptation

In this paper, we seek to develop a versatile test-time adaptation (TTA) objective for a variety of tasks - classification and regression across image-, object-, and pixel-level predictions. We achieve this through a self-bootstrapping scheme that optimizes prediction consistency between the test image (as target) and its deteriorated view. The key challenge lies in devising effective augmentations/deteriorations that: i) preserve the image's geometric information, e.g., object sizes and locations, which is crucial for TTA on object/pixel-level tasks, and ii) provide sufficient learning signals for TTA. To this end, we analyze how common distribution shifts affect the image's information power across spatial frequencies in the Fourier domain, and reveal that low-frequency components carry high power and masking these components supplies more learning signals, while masking high-frequency components can not. In light of this, we randomly mask the low-frequency amplitude of an image in its Fourier domain for augmentation. Meanwhile, we also augment the image with noise injection to compensate for missing learning signals at high frequencies, by enhancing the information power there. Experiments show that, either independently or as a plug-and-play module, our method achieves superior results across classification, segmentation, and 3D monocular detection tasks with both transformer and CNN models.

cs.CV↗

arXiv2025

Continual Optimization with Symmetry Teleportation for Multi-Task Learning

Multi-task learning (MTL) is a widely explored paradigm that enables the simultaneous learning of multiple tasks using a single model. Despite numerous solutions, the key issues of optimization conflict and task imbalance remain under-addressed, limiting performance. Unlike existing optimization-based approaches that typically reweight task losses or gradients to mitigate conflicts or promote progress, we propose a novel approach based on Continual Optimization with Symmetry Teleportation (COST). During MTL optimization, when an optimization conflict arises, we seek an alternative loss-equivalent point on the loss landscape to reduce conflict. Specifically, we utilize a low-rank adapter (LoRA) to facilitate this practical teleportation by designing convergent, loss-invariant objectives. Additionally, we introduce a historical trajectory reuse strategy to continually leverage the benefits of advanced optimizers. Extensive experiments on multiple mainstream datasets demonstrate the effectiveness of our approach. COST is a plug-and-play solution that enhances a wide range of existing MTL methods. When integrated with state-of-the-art methods, COST achieves superior performance.

cs.LG↗

arXiv2025

HDT: Hierarchical Discrete Transformer for Multivariate Time Series Forecasting

Generative models have gained significant attention in multivariate time series forecasting (MTS), particularly due to their ability to generate high-fidelity samples. Forecasting the probability distribution of multivariate time series is a challenging yet practical task. Although some recent attempts have been made to handle this task, two major challenges persist: 1) some existing generative methods underperform in high-dimensional multivariate time series forecasting, which is hard to scale to higher dimensions; 2) the inherent high-dimensional multivariate attributes constrain the forecasting lengths of existing generative models. In this paper, we point out that discrete token representations can model high-dimensional MTS with faster inference time, and forecasting the target with long-term trends of itself can extend the forecasting length with high accuracy. Motivated by this, we propose a vector quantized framework called Hierarchical Discrete Transformer (HDT) that models time series into discrete token representations with l2 normalization enhanced vector quantized strategy, in which we transform the MTS forecasting into discrete tokens generation. To address the limitations of generative models in long-term forecasting, we propose a hierarchical discrete Transformer. This model captures the discrete long-term trend of the target at the low level and leverages this trend as a condition to generate the discrete representation of the target at the high level that introduces the features of the target itself to extend the forecasting length in high-dimensional MTS. Extensive experiments on five popular MTS datasets verify the effectiveness of our proposed method.

cs.LG↗

arXiv2024

Test-Time Model Adaptation with Only Forward Passes

Test-time adaptation has proven effective in adapting a given trained model to unseen test samples with potential distribution shifts. However, in real-world scenarios, models are usually deployed on resource-limited devices, e.g., FPGAs, and are often quantized and hard-coded with non-modifiable parameters for acceleration. In light of this, existing methods are often infeasible since they heavily depend on computation-intensive backpropagation for model updating that may be not supported. To address this, we propose a test-time Forward-Optimization Adaptation (FOA) method. In FOA, we seek to solely learn a newly added prompt (as model's input) via a derivative-free covariance matrix adaptation evolution strategy. To make this strategy work stably under our online unsupervised setting, we devise a novel fitness function by measuring test-training statistic discrepancy and model prediction entropy. Moreover, we design an activation shifting scheme that directly tunes the model activations for shifted test samples, making them align with the source training domain, thereby further enhancing adaptation performance. Without using any backpropagation and altering model weights, FOA runs on quantized 8-bit ViT outperforms gradient-based TENT on full-precision 32-bit ViT, while achieving an up to 24-fold memory reduction on ImageNet-C.

cs.LG↗

arXiv2024

Evolutionary Reinforcement Learning: A Systematic Review and Future Directions

In response to the limitations of reinforcement learning and evolutionary algorithms (EAs) in complex problem-solving, Evolutionary Reinforcement Learning (EvoRL) has emerged as a synergistic solution. EvoRL integrates EAs and reinforcement learning, presenting a promising avenue for training intelligent agents. This systematic review firstly navigates through the technological background of EvoRL, examining the symbiotic relationship between EAs and reinforcement learning algorithms. We then delve into the challenges faced by both EAs and reinforcement learning, exploring their interplay and impact on the efficacy of EvoRL. Furthermore, the review underscores the need for addressing open issues related to scalability, adaptability, sample efficiency, adversarial robustness, ethic and fairness within the current landscape of EvoRL. Finally, we propose future directions for EvoRL, emphasizing research avenues that strive to enhance self-adaptation and self-improvement, generalization, interpretability, explainability, and so on. Serving as a comprehensive resource for researchers and practitioners, this systematic review provides insights into the current state of EvoRL and offers a guide for advancing its capabilities in the ever-evolving landscape of artificial intelligence.

cs.NE↗

arXiv2024

Fine-Tuned Convex Approximations of Probabilistic Reachable Sets under Data-driven Uncertainties

This paper proposes a mechanism to fine-tune convex approximations of probabilistic reachable sets (PRS) of uncertain dynamic systems. We consider the case of unbounded uncertainties, for which it may be impossible to find a bounded reachable set of the system. Instead, we turn to find a PRS that bounds system states with high confidence. Our data-driven approach builds on a kernel density estimator (KDE) accelerated by a fast Fourier transform (FFT), which is customized to model the uncertainties and obtain the PRS efficiently. However, the non-convex shape of the PRS can make it impractical for subsequent optimal designs. Motivated by this, we formulate a mixed integer nonlinear programming (MINLP) problem whose solution result is an optimal $n$ sided convex polygon that approximates the PRS. Leveraging this formulation, we propose a heuristic algorithm to find this convex set efficiently while ensuring accuracy. The algorithm is tested on comprehensive case studies that demonstrate its near-optimality, accuracy, efficiency, and robustness. The benefits of this work pave the way for promising applications to safety-critical, real-time motion planning of uncertain dynamic systems.

cs.RO↗

arXiv2024

Stability Criteria and Calculus Rules via Conic Contingent Coderivatives in Banach Spaces

This paper addresses the study of novel constructions of variational analysis and generalized differentiation that are appropriate for characterizing robust stability properties of constrained set-valued mappings/multifunctions between Banach spaces important in optimization theory and its applications. Our tools of generalized differentiation revolves around the newly introduced concept of $\varepsilon$-regular normal cone to sets and associated coderivative notions for set-valued mappings. Based on these constructions, we establish several characterizations of the central stability notion known as the relative Lipschitz-like property of set-valued mappings in infinite dimensions. Applying a new version of the constrained extremal principle of variational analysis, we develop comprehensive sum and chain rules for our major constructions of conic contingent coderivatives for multifunctions between appropriate classes of Banach spaces.

math.OC↗

arXiv2023

Multi-scale Attention Flow for Probabilistic Time Series Forecasting

The probability prediction of multivariate time series is a notoriously challenging but practical task. On the one hand, the challenge is how to effectively capture the cross-series correlations between interacting time series, to achieve accurate distribution modeling. On the other hand, we should consider how to capture the contextual information within time series more accurately to model multivariate temporal dynamics of time series. In this work, we proposed a novel non-autoregressive deep learning model, called Multi-scale Attention Normalizing Flow(MANF), where we integrate multi-scale attention and relative position information and the multivariate data distribution is represented by the conditioned normalizing flow. Additionally, compared with autoregressive modeling methods, our model avoids the influence of cumulative error and does not increase the time complexity. Extensive experiments demonstrate that our model achieves state-of-the-art performance on many popular multivariate datasets.

cs.LG↗

arXiv2023

A Survey on Reinforcement Learning for Recommender Systems

Recommender systems have been widely applied in different real-life scenarios to help us find useful information. In particular, Reinforcement Learning (RL) based recommender systems have become an emerging research topic in recent years, owing to the interactive nature and autonomous learning ability. Empirical results show that RL-based recommendation methods often surpass most of supervised learning methods. Nevertheless, there are various challenges of applying RL in recommender systems. To understand the challenges and relevant solutions, there should be a reference for researchers and practitioners working on RL-based recommender systems. To this end, we firstly provide a thorough overview, comparisons, and summarization of RL approaches applied in four typical recommendation scenarios, including interactive recommendation, conversational recommendatin, sequential recommendation, and explainable recommendation. Furthermore, we systematically analyze the challenges and relevant solutions on the basis of existing literature. Finally, under discussion for open issues of RL and its limitations of recommender systems, we highlight some potential research directions in this field.

cs.IR↗

arXiv2023

Efficient Training of Large-scale Industrial Fault Diagnostic Models through Federated Opportunistic Block Dropout

Artificial intelligence (AI)-empowered industrial fault diagnostics is important in ensuring the safe operation of industrial applications. Since complex industrial systems often involve multiple industrial plants (possibly belonging to different companies or subsidiaries) with sensitive data collected and stored in a distributed manner, collaborative fault diagnostic model training often needs to leverage federated learning (FL). As the scale of the industrial fault diagnostic models are often large and communication channels in such systems are often not exclusively used for FL model training, existing deployed FL model training frameworks cannot train such models efficiently across multiple institutions. In this paper, we report our experience developing and deploying the Federated Opportunistic Block Dropout (FEDOBD) approach for industrial fault diagnostic model training. By decomposing large-scale models into semantic blocks and enabling FL participants to opportunistically upload selected important blocks in a quantized manner, it significantly reduces the communication overhead while maintaining model performance. Since its deployment in ENN Group in February 2022, FEDOBD has served two coal chemical plants across two cities in China to build industrial fault prediction models. It helped the company reduce the training communication overhead by over 70% compared to its previous AI Engine, while maintaining model performance at over 85% test F1 score. To our knowledge, it is the first successfully deployed dropout-based FL approach.

cs.LG↗

arXiv2022

Probabilistic Guaranteed Path Planning for Safe Urban Air Mobility Using Chance Constrained RRT

Safety is a critical concern for the success of urban air mobility, especially in dynamic and uncertain environments. This paper proposes a path planning algorithm based on RRT in conjunction with chance constraints in the presence of uncertain obstacles. The chance-constrained formulation for Gaussian distributed obstacles is developed by converting the probabilistic constraints to deterministic constraints in terms of distribution parameters. The probabilistic feasible region at every time step can be established through the simulation of the system state and the evaluation of convex constraints. Through establishing chance-constrained RRT, the algorithm not only enjoys the benefits of sampling-based algorithms but also incorporates uncertainty into the formulation. Simulation results demonstrate that the planning for a trajectory connecting the starting and goal point in accordance with the requirement of probabilistic obstacle avoidance can be achieved by the utilization of this algorithm.

cs.RO↗

arXiv2022

Existence of Pareto Solutions for Vector Polynomial Optimization Problems with Constraints

In this paper, we are interested in the existence of Pareto solutions to vector polynomial optimization problems over a basic closed semi-algebraic set. By invoking some powerful tools from real semi-algebraic geometry, we first introduce the concept called {\it tangency varieties}; then we establish connections of the Palais--Smale condition, Cerami condition, {\it M}-tameness, and properness related to the considered problem, in which the condition of Mangasarian--Fromovitz constraint qualification at infinity plays an essential role in deriving these connections. According to the obtained connections, we provide some sufficient conditions for existence of Pareto solutions to the problem in consideration, and we also give some examples to illustrate our main findings.

math.OC↗

arXiv2017

Face Detection using Deep Learning: An Improved Faster RCNN Approach

In this report, we present a new face detection scheme using deep learning and achieve the state-of-the-art detection performance on the well-known FDDB face detetion benchmark evaluation. In particular, we improve the state-of-the-art faster RCNN framework by combining a number of strategies, including feature concatenation, hard negative mining, multi-scale training, model pretraining, and proper calibration of key parameters. As a consequence, the proposed scheme obtained the state-of-the-art face detection performance, making it the best model in terms of ROC curves among all the published methods on the FDDB benchmark.

cs.CV↗

arXiv2015

HDIdx: High-Dimensional Indexing for Efficient Approximate Nearest Neighbor Search

Fast Nearest Neighbor (NN) search is a fundamental challenge in large-scale data processing and analytics, particularly for analyzing multimedia contents which are often of high dimensionality. Instead of using exact NN search, extensive research efforts have been focusing on approximate NN search algorithms. In this work, we present "HDIdx", an efficient high-dimensional indexing library for fast approximate NN search, which is open-source and written in Python. It offers a family of state-of-the-art algorithms that convert input high-dimensional vectors into compact binary codes, making them very efficient and scalable for NN search with very low space complexity.

cs.IR↗

arXiv2015

A Framework of Sparse Online Learning and Its Applications

The amount of data in our society has been exploding in the era of big data today. In this paper, we address several open challenges of big data stream classification, including high volume, high velocity, high dimensionality, high sparsity, and high class-imbalance. Many existing studies in data mining literature solve data stream classification tasks in a batch learning setting, which suffers from poor efficiency and scalability when dealing with big data. To overcome the limitations, this paper investigates an online learning framework for big data stream classification tasks. Unlike some existing online data stream classification techniques that are often based on first-order online learning, we propose a framework of Sparse Online Classification (SOC) for data stream classification, which includes some state-of-the-art first-order sparse online learning algorithms as special cases and allows us to derive a new effective second-order online learning algorithm for data stream classification. In addition, we also propose a new cost-sensitive sparse online learning algorithm by extending the framework with application to tackle online anomaly detection tasks where class distribution of data could be very imbalanced. We also analyze the theoretical bounds of the proposed method, and finally conduct an extensive set of experiments, in which encouraging results validate the efficacy of the proposed algorithms in comparison to a family of state-of-the-art techniques on a variety of data stream classification tasks.

cs.LG↗

arXiv2012

Fast Bounded Online Gradient Descent Algorithms for Scalable Kernel-Based Online Learning

Kernel-based online learning has often shown state-of-the-art performance for many online learning tasks. It, however, suffers from a major shortcoming, that is, the unbounded number of support vectors, making it non-scalable and unsuitable for applications with large-scale datasets. In this work, we study the problem of bounded kernel-based online learning that aims to constrain the number of support vectors by a predefined budget. Although several algorithms have been proposed in literature, they are neither computationally efficient due to their intensive budget maintenance strategy nor effective due to the use of simple Perceptron algorithm. To overcome these limitations, we propose a framework for bounded kernel-based online learning based on an online gradient descent approach. We propose two efficient algorithms of bounded online gradient descent (BOGD) for scalable kernel-based online learning: (i) BOGD by maintaining support vectors using uniform sampling, and (ii) BOGD++ by maintaining support vectors using non-uniform sampling. We present theoretical analysis of regret bound for both algorithms, and found promising empirical performance in terms of both efficacy and efficiency by comparing them to several well-known algorithms for bounded kernel-based online learning on large-scale datasets.

cs.LG↗