Distributed Learning Research Articles

Instance segmentation on distributed deep learning big data cluster

AbstractDistributed deep learning is a promising approach for training and deploying large and complex deep learning models. This paper presents a comprehensive workflow for deploying and optimizing the YOLACT instance segmentation model as on big data clusters. OpenVINO, a toolkit known for its high-speed data processing and ability to optimize deep learning models for deployment on a variety of devices, was used to optimize the YOLACT model. The model is then run on a big data cluster using BigDL, a distributed deep learning library for Apache Spark. BigDL provides a high-level programming interface for defining and training deep neural networks, making it suitable for large-scale deep learning applications. In distributed deep learning, input data is divided and distributed across multiple machines for parallel processing. This approach offers several advantages, including the ability to handle very large data that can be stored in a distributed manner, scalability to decrease processing time by increasing the number of workers, and fault tolerance. The proposed workflow was evaluated on virtual machines and Azure Databricks, a cloud-based platform for big data analytics. The results indicated that the workflow can scale to large datasets and deliver high performance on Azure Databricks. This study explores the benefits and challenges of using distributed deep learning on big data clusters for instance segmentation. Popular distributed deep learning frameworks are discussed, and BigDL is chosen. Overall, this study highlights the practicality of distributed deep learning for deploying and scaling sophisticated deep learning models on big data clusters.

Improving Risk-averse Distributional Reinforcement Learning with Adaptive Risk-seeking Exploration

In autonomous driving, reinforcement learning’s interactive learning capabilities have garnered interest. Among these, risk-averse distributional reinforcement learning has demonstrated promise in situations where safety is crucial. On the other hand, risk-averse policies degrade training performance by discouraging exploration. To provide an adaptive exploration policy, this work offers a novel framework for risk-averse distributional reinforcement learning. The proposed method is evaluated in a highly uncertain safety-critical scenario, motion planning for pedestrian avoidance in the face of uncertain interactions. The simulation results demonstrate improved safety, achieving training flexibility between exploration and exploitation.

Spectral Tensor Layers for Communication-Free Distributed Deep Learning.

In this article, we propose a novel spectral tensor layer for communication-free distributed deep learning. The overall framework is as follows: first, we represent the data in tensor form (instead of vector form) and replace the matrix product in conventional neural networks with the tensor product, which in effect imposes certain transformed-induced structure on the original weight matrices, e.g., a block-circulant structure; then, we apply a linear transform along a certain dimension to split the original dataset into multiple spectral subdatasets; as a result, the proposed spectral tensor network consists of parallel branches where each branch is a conventional neural network trained on a spectral subdataset with ZERO communication cost. The parallel branches are directly ensembled (i.e., the weighted sum of their outputs) to generate an overall network with substantially stronger generalization capability than that of each branch. Moreover, the proposed method enjoys a byproduct of decentralization gain in terms of memory and computation, compared with traditional networks. It is a natural yet elegant solution for heterogeneous data in federated learning (FL), where data at different nodes have different resolutions. Finally, we evaluate the proposed spectral tensor networks on the MNIST, CIFAR-10, ImageNet-1K, and ImageNet-21K datasets, respectively, to verify that they simultaneously achieve communication-free distributed learning, distributed storage reduction, parallel computation speedup, and learning with multiresolution data.

Noise Distribution Decomposition based Multi-Agent Distributional Reinforcement Learning

Noise Distribution Decomposition based Multi-Agent Distributional Reinforcement Learning

Privacy-Preserving Distributed Transfer Learning and its Application in Intelligent Transportation

With the rapid development of intelligent transportation systems (ITS), more and more intelligent applications for ITS have received widespread attention, such as the vehicle detection, inference of typical routes, and traffic forecasting. In these applications, deep learning is widely used as a key artificial intelligence technology. However, most ITS providers fail to collect enough labeled traffic data for model training. As a complement to deep learning, transfer learning is an effective way to solve the scarcity of labeled data, which can transfer knowledge from labeled datasets to unlabeled datasets, thus improving the accuracy of prediction and classification. Nevertheless, when the labeled dataset and the unlabeled dataset are held by different entities, it is still unrealistic for two mutually distrustful entities to cooperate in transfer learning regarding data security and privacy preservation. Although some existing works provide privacy-preserving transfer learning methods, such methods fail to apply to traffic data with high sample dimensions due to their high computational cost and round complexity. To address this problem, we design an efficient privacy-preserving distributed transfer learning protocol, which is appropriate for traffic data. Compared to existing works, our protocol addresses the privacy-preserving problem of transfer learning for traffic data with high sample dimensions. In addition, our protocol has fewer interaction rounds and can be proved in the semi-honest model. Finally, we validate the effectiveness, efficiency and security of the proposed protocol via experiments. Furthermore, we show the application of the proposed protocol in intelligent transportation systems.

Investigation on Preserving Privacy of Electronic Medical Record using Split Learning

Artificial Intelligence is deployed in multiple areas, including healthcare. Utmost research is done in AI enabled healthcare industry because of the demands like accurate result, data security, exact prediction, huge volume of data, etc. In conventional deep learning models, the training happens with the dataset that are stored in a single device. This requires a huge storage space and highly efficient machines to train the data. Usage of big data, demands for innovative models that can be deployed and used in confined storage. Split learning is one such collaborative distributed deep learning model that allows the data to be stored in a split fashion. Split learning supports desirable features like less storage, more privacy to raw data, ability to work with resource constraints, etc., making it suitable for storing electronic medical record of patients. This paper discusses the advantages of using split learning for healthcare, the possible configurations of split learning that supports data privacy in healthcare and finally discusses the open research challenges in implementing split learning for healthcare.

Distributional Policy Gradient With Distributional Value Function.

In this article, we propose a distributional policy-gradient method based on distributional reinforcement learning (RL) and policy gradient. Conventional RL algorithms typically estimate the expectation of return, given a state-action pair. Furthermore, distributional RL algorithms consider the return as a random variable and estimate the return distribution that can characterize the probability of different returns resulted by environmental uncertainties. Thus, the return distribution provides more valuable information than its expectation, leading to superior policies in general. Although distributional RL has been investigated widely in value-based RL methods, very few policy-gradient methods take advantage of distributional RL. To bridge this research gap, we propose a distributional policy-gradient method by introducing a distributional value function to the policy gradient (DVDPG). We estimate the distribution of policy gradient instead of the expectation estimated in conventional policy-gradient RL methods. Furthermore, we propose two policy-gradient value sampling mechanisms to do policy improvement. First, we propose a distribution-probability-sampling method that samples the policy-gradient value according to the quantile probability of return distribution. Second, a uniform sample mechanism is proposed. With our sample mechanisms, the proposed distributional policy-gradient method enhances the stochasticity of the policy gradient, improving the exploration efficiency and benefiting to avoid falling into local optimal solutions. In sparse-reward tasks, the distribution-probability-sampling method outperforms the uniform sample mechanism. In dense-reward tasks, the two sample mechanisms perform similarly. Moreover, we show that the conventional policy-gradient method is a special case of the proposed method. Experimental results on various sparse-reward and dense-reward OpenAI-gym tasks illustrate the efficiency of the proposed method, outperforming baselines in almost environments.

Hedging Beyond the Mean: A Distributional Reinforcement Learning Perspective for Hedging Portfolios with Structured Products

Hedging Beyond the Mean: A Distributional Reinforcement Learning Perspective for Hedging Portfolios with Structured Products

A Hierarchical Distributed Data-Driven Adaptive Learning Control for Nonaffine Nonlinear MASs.

This article designs a new hierarchical distributed data-driven adaptive learning control algorithm to accomplish the leader-following tracking control objective for nonaffine nonlinear multiagent systems (MASs). The proposed hierarchical control structure is composed of a distributed observer and a decentralized data-driven adaptive learning controller. Considering that some followers cannot directly receive information from the leader, a distributed observer is designed to estimate the information of the leader. Based on this, a decentralized data-driven adaptive learning controller is further devised to enable the follower to track the estimated information of the leader, where the model parameter learning algorithm is developed to capture the dynamic characteristics of the original system. One advantage of the developed hierarchical control learning algorithm is that neither the leader's system model nor the follower's system model is needed. The other one is the elimination of the noncausal problem without the additional assumption. Simulation results exemplify the merits of the theoretical results by comparisons.

BIRD+: Design of a Lightweight Communication Compressor for Resource-Constrained Distribution Learning Platforms

BIRD+: Design of a Lightweight Communication Compressor for Resource-Constrained Distribution Learning Platforms

An Efficient Matching Game Approach to Association Formation in UAV-Enabled Hierarchical Distributed Learning.

Distributed machine learning has emerged as a promising data processing technology for next-generation communication systems. It leverages the computational capabilities of local nodes to efficiently handle large datasets, creating highly accurate data-driven models for analysis and prediction purposes. However, the performance of distributed machine learning can be significantly hampered by communication bottlenecks and node dropouts. In this article, a novel unmanned aerial vehicle (UAV)-enabled hierarchical distributed learning architecture is proposed to support machine learning applications, e.g., regional monitoring. Multiple UAV receivers (URs) are introduced as wireless relays to improve the communication between the UAV transmitters (UTs) and the cloud server. Our objective is to identify the optimal UT-UR association to maximize the social welfare of the network, which is distinctly different from the existing works that focus on the unilateral profit-maximizing problem. We formulate a two-side many-to-one matching game to model the UT-UR association problem, and a two-phase many-to-one matching algorithm is designed to identify the stable matching. The validity of our proposed scheme is verified through in-depth numerical simulations.

UDSem: A Unified Distributed Learning Framework for Semantic Communications Over Wireless Networks

End-to-end semantic communications (ESC) rely on deep neural networks (DNN) to boost the communication efficiency by only transmitting the semantics of data. However, ESC is data-hungry for training millions of neural parameters, while uploading massive data generated by users/devices to a central server is usually impractical due to privacy issues, regulatory guidelines and network congestion, etc. Inspired by federated learning, where each device contributes to the parameters' update by independently computing the gradient based on its local data, we present UDSem, a unified distributed learning framework of semantic communications for texts and images over wireless networks. The key ingredients of our UDSem include 1) a flexible learning mechanism that can be tailored to a device based on its computing capabilities, 2) an efficient learning approach that splits the system into multiple modules and updates the parameters independently both on clients and the central server, 3) a mixed aggregation method that globally updates the neural parameters for each module of the ESC system. Equipped with these ingredients, our proposed UDSem is able to properly fill the gap between the training efficiency of ESC systems and the privacy requirement of users' data. Experimental results over two public benchmarks show the superiority of our UDSem in terms of convergence speed and semantic interpretation, potentially paving the way for distributed semantic communications in future wireless networks.

Domain Adaptation for Label Distribution Learning

Domain Adaptation for Label Distribution Learning

Learning Temporal Distribution and Spatial Correlation Toward Universal Moving Object Segmentation.

The goal of moving object segmentation is separating moving objects from stationary backgrounds in videos. One major challenge in this problem is how to develop a universal model for videos from various natural scenes since previous methods are often effective only in specific scenes. In this paper, we propose a method called Learning Temporal Distribution and Spatial Correlation (LTS) that has the potential to be a general solution for universal moving object segmentation. In the proposed approach, the distribution from temporal pixels is first learned by our Defect Iterative Distribution Learning (DIDL) network for a scene-independent segmentation. Notably, the DIDL network incorporates the use of an improved product distribution layer that we have newly derived. Then, the Stochastic Bayesian Refinement (SBR) Network, which learns the spatial correlation, is proposed to improve the binary mask generated by the DIDL network. Benefiting from the scene independence of the temporal distribution and the accuracy improvement resulting from the spatial correlation, the proposed approach performs well for almost all videos from diverse and complex natural scenes with fixed parameters. Comprehensive experiments on standard datasets including LASIESTA, CDNet2014, BMC, SBMI2015 and 128 real world videos demonstrate the superiority of proposed approach compared to state-of-the-art methods with or without the use of deep learning networks. To the best of our knowledge, this work has high potential to be a general solution for moving object segmentation in real world environments. The code and real-world videos can be found on GitHub https://github.com/guanfangdong/LTS-UniverisalMOS.

D3NN: Data-Driven Dual-level Neural Network Tool for Dynamical System Modeling and Abstraction

This paper presents D3NN, a neural network learning tool for dual-level dynamic system modeling. This tool aims to model the dynamic system under a dual-level modeling framework, i.e., modeling the system dynamics through a computationally efficient and precise low-level neural hybrid system and an interpretable high-level transition system abstraction. Given samples of the dynamic system and user specifications, the proposed tool can generate a low-level distributive learning framework consisting of shallow neural network approximations for local dynamics and its high-level transition system abstraction that allows Computational Tree Logic (CTL) formulae verification.

A Study of Federated Learning with Internet of Things for Data Privacy and Security using Privacy Preserving Techniques

Abstract: Privacy leakage that occurs when many IoT devices are utilized for training centralized models, a new distributed learning framework known as federated learning was created, where devices train models together while keeping their private datasets local. In a federated learning setup, a central aggregator coordinates the efforts of several clients working together to solve machine learning issues. The privacy of each device's data is protected by this setup's decentralized training data. Federated learning reduces traditional centralized machine learning systems' systemic privacy issues and costs by emphasizing local processing and model transfer. Client information is stored locally and cannot be copied or shared. By utilizing a centralized server, federated learning enables each participant's device to collect data locally for training purposes before sending the resulting model to the server for aggregate and subsequent distribution. As a means of providing a comprehensive review and encouraging further research into the topic, we introduce the works of federated learning from five different vantage points: data partitioning, privacy method, machine learning model, communication architecture, and systems heterogeneity. Then, we organize the issues plaguing federated learning today and the potential avenues for a prospective study. Finally, we provide a brief overview of the features of existing federated knowledge and discuss how it is currently being used in the field.

Why Batch Normalization Damage Federated Learning on Non-IID Data?

As a promising distributed learning paradigm, federated learning (FL) involves training deep neural network (DNN) models at the network edge while protecting the privacy of the edge clients. To train a large-scale DNN model, batch normalization (BN) has been regarded as a simple and effective means to accelerate the training and improve the generalization capability. However, recent findings indicate that BN can significantly impair the performance of FL in the presence of non-i.i.d. data. While several FL algorithms have been proposed to address this issue, their performance still falls significantly when compared to the centralized scheme. Furthermore, none of them have provided a theoretical explanation of how the BN damages the FL convergence. In this article, we present the first convergence analysis to show that under the non-i.i.d. data, the mismatch between the local and global statistical parameters in BN causes the gradient deviation between the local and global models, which, as a result, slows down and biases the FL convergence. In view of this, we develop a new FL algorithm that is tailored to BN, called FedTAN, which is capable of achieving robust FL performance under a variety of data distributions via iterative layer-wise parameter aggregation. Comprehensive experimental results demonstrate the superiority of the proposed FedTAN over existing baselines for training BN-based DNN models.

Uncertainty Modeling for Gaze Estimation.

Gaze estimation is an important fundamental task in computer vision and medical research. Existing works have explored various effective paradigms and modules for precisely predicting eye gazes. However, the uncertainty for gaze estimation, e.g., input uncertainty and annotation uncertainty, have been neglected in previous research. Existing models use a deterministic function to estimate the gaze, which cannot reflect the actual situation in gaze estimation. To address this issue, we propose a probabilistic framework for gaze estimation by modeling the input uncertainty and annotation uncertainty. We first utilize probabilistic embeddings to model the input uncertainty, representing the input image as a Gaussian distribution in the embedding space. Based on the input uncertainty modeling, we give an instance-wise uncertainty estimation to measure the confidence of prediction results, which is critical in practical applications. Then, we propose a new label distribution learning method, probabilistic annotations, to model the annotation uncertainty, representing the raw hard labels as Gaussian distributions. In addition, we develop an Embedding Distribution Smoothing (EDS) module and a hard example mining method to improve the consistency between embedding distribution and label distribution. We conduct extensive experiments, demonstrating that the proposed approach achieves significant improvements over baseline and state-of-the-art methods on two widely used benchmark datasets, GazeCapture and MPIIFaceGaze, as well as our collected dataset using mobile devices.

Allosteric Feature Collaboration for Model-Heterogeneous Federated Learning.

Although federated learning (FL) has achieved outstanding results in privacy-preserved distributed learning, the setting of model homogeneity among clients restricts its wide application in practice. This article investigates a more general case, namely, model-heterogeneous FL (M-hete FL), where client models are independently designed and can be structurally heterogeneous. M-hete FL faces new challenges in collaborative learning because the parameters of heterogeneous models could not be directly aggregated. In this article, we propose a novel allosteric feature collaboration (AlFeCo) method, which interchanges knowledge across clients and collaboratively updates heterogeneous models on the server. Specifically, an allosteric feature generator is developed to reveal task-relevant information from multiple client models. The revealed information is stored in the client-shared and client-specific codes. We exchange client-specific codes across clients to facilitate knowledge interchange and generate allosteric features that are dimensionally variable for model updates. To promote information communication between different clients, a dual-path (model-model and model-prediction) communication mechanism is designed to supervise the collaborative model updates using the allosteric features. Client models are fully communicated through the knowledge interchange between models and between models and predictions. We further provide theoretical evidence and convergence analysis to support the effectiveness of AlFeCo in M-hete FL. The experimental results show that the proposed AlFeCo method not only performs well on classical FL benchmarks but also is effective in model-heterogeneous federated antispoofing. Our codes are publicly available at https://github.com/ybaoyao/AlFeCo.

Auditable Federated Learning With Byzantine Robustness

Machine learning (ML) has led to disruptive innovations in many fields, such as medical diagnoses. A key enabler for ML is large training data, but existing data, such as medical data, are not fully exploited by ML because of data silos and privacy concerns. Federated learning (FL) is a promising distributed learning paradigm to address this problem. On the other hand, existing FL approaches are vulnerable to poisoning attacks or privacy leakage from a malicious aggregator or client. This article proposes an auditable FL scheme with Byzantine robustness against the aggregator and client: The aggregator is malicious but available, and the client could perform poisoning attacks. First, the Pedersen commitment scheme (PCS) for homomorphic encryption was applied to preserve privacy and for commitments to the FL process to achieve auditability. The auditability enables clients to verify the correctness and consistency of the entire FL process and to identify parties that misbehave. Second, an efficient technique of divide and conquer was designed based on PCS to allow parties to cooperate and securely aggregate gradients to defend against poisoning attacks. This technique enables clients to share no common secret key and cooperate to decrypt ciphertext, guaranteeing a client’s privacy even if some other clients are corrupted by adversaries. This technique was optimized to tolerate the dropout of clients. This article reports a formal analysis concerning privacy, efficiency, and auditability against malicious participants. Extensive experiments on various benchmark datasets show that the scheme is robust with high model accuracy against poisoning attacks.