arXiv subjects

Jihong Park

Publications and source records attributed to Jihong Park.

At least 109 records · Page 6Linked to original sources

arXiv2022

Quantum Multi-Agent Reinforcement Learning via Variational Quantum Circuit Design

In recent years, quantum computing (QC) has been getting a lot of attention from industry and academia. Especially, among various QC research topics, variational quantum circuit (VQC) enables quantum deep reinforcement learning (QRL). Many studies of QRL have shown that the QRL is superior to the classical reinforcement learning (RL) methods under the constraints of the number of training parameters. This paper extends and demonstrates the QRL to quantum multi-agent RL (QMARL). However, the extension of QRL to QMARL is not straightforward due to the challenge of the noise intermediate-scale quantum (NISQ) and the non-stationary properties in classical multi-agent RL (MARL). Therefore, this paper proposes the centralized training and decentralized execution (CTDE) QMARL framework by designing novel VQCs for the framework to cope with these issues. To corroborate the QMARL framework, this paper conducts the QMARL demonstration in a single-hop environment where edge agents offload packets to clouds. The extensive demonstration shows that the proposed QMARL framework enhances 57.7% of total reward than classical frameworks.

quant-ph↗

arXiv2022

A Unified View on Semantic Information and Communication: A Probabilistic Logic Approach

This article aims to provide a unified and technical approach to semantic information, communication, and their interplay through the lens of probabilistic logic. To this end, on top of the existing technical communication (TC) layer, we additionally introduce a semantic communication (SC) layer that exchanges logically meaningful clauses in knowledge bases. To make these SC and TC layers interact, we propose various measures based on the entropy of a clause in a knowledge base. These measures allow us to delineate various technical issues on SC such as a message selection problem for improving the knowledge at a receiver. Extending this, we showcase selected examples in which SC and TC layers interact with each other while taking into account constraints on physical channels.

cs.IT↗

arXiv2021

Attention Based Communication and Control for Multi-UAV Path Planning

Inspired by the multi-head attention (MHA) mechanism in natural language processing, this letter proposes an iterative single-head attention (ISHA) mechanism for multi-UAV path planning. The ISHA mechanism is run by a communication helper collecting the state embeddings of UAVs and distributing an attention score vector to each UAV. The attention scores computed by ISHA identify how many interactions with other UAVs should be considered in each UAV's control decision-making. Simulation results corroborate that the ISHA-based communication and control framework achieves faster travel with lower inter-UAV collision risks than an MHA-aided baseline, particularly under limited communication resources.

cs.LG↗

arXiv2021

Server-Side Local Gradient Averaging and Learning Rate Acceleration for Scalable Split Learning

In recent years, there have been great advances in the field of decentralized learning with private data. Federated learning (FL) and split learning (SL) are two spearheads possessing their pros and cons, and are suited for many user clients and large models, respectively. To enjoy both benefits, hybrid approaches such as SplitFed have emerged of late, yet their fundamentals have still been illusive. In this work, we first identify the fundamental bottlenecks of SL, and thereby propose a scalable SL framework, coined SGLR. The server under SGLR broadcasts a common gradient averaged at the split-layer, emulating FL without any additional communication across clients as opposed to SplitFed. Meanwhile, SGLR splits the learning rate into its server-side and client-side rates, and separately adjusts them to support many clients in parallel. Simulation results corroborate that SGLR achieves higher accuracy than other baseline SL methods including SplitFed, which is even on par with FL consuming higher energy and communication costs. As a secondary result, we observe greater reduction in leakage of sensitive information via mutual information using SLGR over the baselines.

cs.LG↗

arXiv2021

Joint Superposition Coding and Training for Federated Learning over Multi-Width Neural Networks

This paper aims to integrate two synergetic technologies, federated learning (FL) and width-adjustable slimmable neural network (SNN) architectures. FL preserves data privacy by exchanging the locally trained models of mobile devices. By adopting SNNs as local models, FL can flexibly cope with the time-varying energy capacities of mobile devices. Combining FL and SNNs is however non-trivial, particularly under wireless connections with time-varying channel conditions. Furthermore, existing multi-width SNN training algorithms are sensitive to the data distributions across devices, so are ill-suited to FL. Motivated by this, we propose a communication and energy-efficient SNN-based FL (named SlimFL) that jointly utilizes superposition coding (SC) for global model aggregation and superposition training (ST) for updating local models. By applying SC, SlimFL exchanges the superposition of multiple width configurations that are decoded as many as possible for a given communication throughput. Leveraging ST, SlimFL aligns the forward propagation of different width configurations, while avoiding the inter-width interference during backpropagation. We formally prove the convergence of SlimFL. The result reveals that SlimFL is not only communication-efficient but also can counteract non-IID data distributions and poor channel conditions, which is also corroborated by simulations.

cs.LG↗

arXiv2021

Communication and Energy Efficient Slimmable Federated Learning via Superposition Coding and Successive Decoding

Mobile devices are indispensable sources of big data. Federated learning (FL) has a great potential in exploiting these private data by exchanging locally trained models instead of their raw data. However, mobile devices are often energy limited and wirelessly connected, and FL cannot cope flexibly with their heterogeneous and time-varying energy capacity and communication throughput, limiting the adoption. Motivated by these issues, we propose a novel energy and communication efficient FL framework, coined SlimFL. To resolve the heterogeneous energy capacity problem, each device in SlimFL runs a width-adjustable slimmable neural network (SNN). To address the heterogeneous communication throughput problem, each full-width (1.0x) SNN model and its half-width ($0.5$x) model are superposition-coded before transmission, and successively decoded after reception as the 0.5x or $1.0$x model depending on the channel quality. Simulation results show that SlimFL can simultaneously train both $0.5$x and $1.0$x models with reasonable accuracy and convergence speed, compared to its vanilla FL counterpart separately training the two models using $2$x more communication resources. Surprisingly, SlimFL achieves even higher accuracy with lower energy footprints than vanilla FL for poor channels and non-IID data distributions, under which vanilla FL converges slowly.

cs.LG↗

arXiv2021

Learning Emergent Random Access Protocol for LEO Satellite Networks

A mega-constellation of low-altitude earth orbit (LEO) satellites (SATs) are envisaged to provide a global coverage SAT network in beyond fifth-generation (5G) cellular systems. LEO SAT networks exhibit extremely long link distances of many users under time-varying SAT network topology. This makes existing multiple access protocols, such as random access channel (RACH) based cellular protocol designed for fixed terrestrial network topology, ill-suited. To overcome this issue, in this paper, we propose a novel grant-free random access solution for LEO SAT networks, dubbed emergent random access channel protocol (eRACH). In stark contrast to existing model-based and standardized protocols, eRACH is a model-free approach that emerges through interaction with the non-stationary network environment, using multi-agent deep reinforcement learning (MADRL). Furthermore, by exploiting known SAT orbiting patterns, eRACH does not require central coordination or additional communication across users, while training convergence is stabilized through the regular orbiting patterns. Compared to RACH, we show from various simulations that our proposed eRACH yields 54.6% higher average network throughput with around two times lower average access delay while achieving 0.989 Jain's fairness index.

cs.IT↗

arXiv2021

Attention-based Reinforcement Learning for Real-Time UAV Semantic Communication

In this article, we study the problem of air-to-ground ultra-reliable and low-latency communication (URLLC) for a moving ground user. This is done by controlling multiple unmanned aerial vehicles (UAVs) in real time while avoiding inter-UAV collisions. To this end, we propose a novel multi-agent deep reinforcement learning (MADRL) framework, coined a graph attention exchange network (GAXNet). In GAXNet, each UAV constructs an attention graph locally measuring the level of attention to its neighboring UAVs, while exchanging the attention weights with other UAVs so as to reduce the attention mismatch between them. Simulation results corroborates that GAXNet achieves up to 4.5x higher rewards during training. At execution, without incurring inter-UAV collisions, GAXNet achieves 6.5x lower latency with the target 0.0000001 error rate, compared to a state-of-the-art baseline framework.

cs.MA↗

arXiv2021

Robust Reconfigurable Intelligent Surfaces via Invariant Risk and Causal Representations

In this paper, the problem of robust reconfigurable intelligent surface (RIS) system design under changes in data distributions is investigated. Using the notion of invariant risk minimization (IRM), an invariant causal representation across multiple environments is used such that the predictor is simultaneously optimal for each environment. A neural network-based solution is adopted to seek the predictor and its performance is validated via simulations against an empirical risk minimization-based design. Results show that leveraging invariance yields more robustness against unseen and out-of-distribution testing environments.

cs.LG↗

arXiv2021

AirMixML: Over-the-Air Data Mixup for Inherently Privacy-Preserving Edge Machine Learning

Wireless channels can be inherently privacy-preserving by distorting the received signals due to channel noise, and superpositioning multiple signals over-the-air. By harnessing these natural distortions and superpositions by wireless channels, we propose a novel privacy-preserving machine learning (ML) framework at the network edge, coined over-the-air mixup ML (AirMixML). In AirMixML, multiple workers transmit analog-modulated signals of their private data samples to an edge server who trains an ML model using the received noisy-and superpositioned samples. AirMixML coincides with model training using mixup data augmentation achieving comparable accuracy to that with raw data samples. From a privacy perspective, AirMixML is a differentially private (DP) mechanism limiting the disclosure of each worker's private sample information at the server, while the worker's transmit power determines the privacy disclosure level. To this end, we develop a fractional channel-inversion power control (PC) method, α-Dirichlet mixup PC (DirMix(α)-PC), wherein for a given global power scaling factor after channel inversion, each worker's local power contribution to the superpositioned signal is controlled by the Dirichlet dispersion ratio α. Mathematically, we derive a closed-form expression clarifying the relationship between the local and global PC factors to guarantee a target DP level. By simulations, we provide DirMix(α)-PC design guidelines to improve accuracy, privacy, and energy-efficiency. Finally, AirMixML with DirMix(α)-PC is shown to achieve reasonable accuracy compared to a privacy-violating baseline with neither superposition nor PC.

cs.NI↗

arXiv2021

Communication-Efficient and Personalized Federated Lottery Ticket Learning

The lottery ticket hypothesis (LTH) claims that a deep neural network (i.e., ground network) contains a number of subnetworks (i.e., winning tickets), each of which exhibiting identically accurate inference capability as that of the ground network. Federated learning (FL) has recently been applied in LotteryFL to discover such winning tickets in a distributed way, showing higher accuracy multi-task learning than Vanilla FL. Nonetheless, LotteryFL relies on unicast transmission on the downlink, and ignores mitigating stragglers, questioning scalability. Motivated by this, in this article we propose a personalized and communication-efficient federated lottery ticket learning algorithm, coined CELL, which exploits downlink broadcast for communication efficiency. Furthermore, it utilizes a novel user grouping method, thereby alternating between FL and lottery learning to mitigate stragglers. Numerical simulations validate that CELL achieves up to 3.6% higher personalized task classification accuracy with 4.3x smaller total communication cost until convergence under the CIFAR-10 dataset.

cs.LG↗

arXiv2021

Split Learning Meets Koopman Theory for Wireless Remote Monitoring and Prediction

Remote state monitoring over wireless is envisaged to play a pivotal role in enabling beyond 5G applications ranging from remote drone control to remote surgery. One key challenge is to identify the system dynamics that is non-linear with a large dimensional state. To obviate this issue, in this article we propose to train an autoencoder whose encoder and decoder are split and stored at a state sensor and its remote observer, respectively. This autoencoder not only decreases the remote monitoring payload size by reducing the state representation dimension, but also learns the system dynamics by lifting it via a Koopman operator, thereby allowing the observer to locally predict future states after training convergence. Numerical results under a non-linear cart-pole environment demonstrate that the proposed split learning of a Koopman autoencoder can locally predict future states, and the prediction accuracy increases with the representation dimension and transmission power.

cs.LG↗

arXiv2021

Robustness and Diversity Seeking Data-Free Knowledge Distillation

Knowledge distillation (KD) has enabled remarkable progress in model compression and knowledge transfer. However, KD requires a large volume of original data or their representation statistics that are not usually available in practice. Data-free KD has recently been proposed to resolve this problem, wherein teacher and student models are fed by a synthetic sample generator trained from the teacher. Nonetheless, existing data-free KD methods rely on fine-tuning of weights to balance multiple losses, and ignore the diversity of generated samples, resulting in limited accuracy and robustness. To overcome this challenge, we propose robustness and diversity seeking data-free KD (RDSKD) in this paper. The generator loss function is crafted to produce samples with high authenticity, class diversity, and inter-sample diversity. Without real data, the objectives of seeking high sample authenticity and class diversity often conflict with each other, causing frequent loss fluctuations. We mitigate this by exponentially penalizing loss increments. With MNIST, CIFAR-10, and SVHN datasets, our experiments show that RDSKD achieves higher accuracy with more robustness over different hyperparameter settings, compared to other data-free KD methods such as DAFL, MSKD, ZSKD, and DeepInversion.

cs.LG↗

arXiv2021

Predictive Control and Communication Co-Design via Two-Way Gaussian Process Regression and AoI-Aware Scheduling

This article studies the joint problem of uplink-downlink scheduling and power allocation for controlling a large number of actuators that upload their states to remote controllers and download control actions over wireless links. To overcome the lack of wireless resources, we propose a machine learning-based solution, where only a fraction of actuators is controlled, while the rest of the actuators are actuated by locally predicting the missing state and/or action information using the previous uplink and/or downlink receptions via a Gaussian process regression (GPR). This GPR prediction credibility is determined using the age-of-information (AoI) of the latest reception. Moreover, the successful reception is affected by the transmission power, mandating a co-design of the communication and control operations. To this end, we formulate a network-wide minimization problem of the average AoI and transmission power under communication reliability and control stability constraints. To solve the problem, we propose a dynamic control algorithm using the Lyapunov drift-plus-penalty optimization framework. Numerical results corroborate that the proposed algorithm can stably control $2$x more number of actuators compared to an event-triggered scheduling baseline with Kalman filtering and frequency division multiple access, which is $18$x larger than a round-robin scheduling baseline.

cs.IT↗

arXiv2021

Communication Efficient Distributed Learning with Censored, Quantized, and Generalized Group ADMM

In this paper, we propose a communication-efficiently decentralized machine learning framework that solves a consensus optimization problem defined over a network of inter-connected workers. The proposed algorithm, Censored and Quantized Generalized GADMM (CQ-GGADMM), leverages the worker grouping and decentralized learning ideas of Group Alternating Direction Method of Multipliers (GADMM), and pushes the frontier in communication efficiency by extending its applicability to generalized network topologies, while incorporating link censoring for negligible updates after quantization. We theoretically prove that CQ-GGADMM achieves the linear convergence rate when the local objective functions are strongly convex under some mild assumptions. Numerical simulations corroborate that CQ-GGADMM exhibits higher communication efficiency in terms of the number of communication rounds and transmit energy consumption without compromising the accuracy and convergence speed, compared to the censored decentralized ADMM, and the worker grouping method of GADMM.

cs.LG↗

arXiv2021

Robust Blockchained Federated Learning with Model Validation and Proof-of-Stake Inspired Consensus

Federated learning (FL) is a promising distributed learning solution that only exchanges model parameters without revealing raw data. However, the centralized architecture of FL is vulnerable to the single point of failure. In addition, FL does not examine the legitimacy of local models, so even a small fraction of malicious devices can disrupt global training. To resolve these robustness issues of FL, in this paper, we propose a blockchain-based decentralized FL framework, termed VBFL, by exploiting two mechanisms in a blockchained architecture. First, we introduced a novel decentralized validation mechanism such that the legitimacy of local model updates is examined by individual validators. Second, we designed a dedicated proof-of-stake consensus mechanism where stake is more frequently rewarded to honest devices, which protects the legitimate local model updates by increasing their chances of dictating the blocks appended to the blockchain. Together, these solutions promote more federation within legitimate devices, enabling robust FL. Our emulation results of the MNIST classification corroborate that with 15% of malicious devices, VBFL achieves 87% accuracy, which is 7.4x higher than Vanilla FL.

cs.LG↗

arXiv2021

RIS-Assisted Coverage Enhancement in Millimeter-Wave Cellular Networks

The use of millimeter-wave (mmWave) bandwidth is one key enabler to achieve the high data rates in the fifth-generation (5G) cellular systems. However, mmWave signals suffer from significant path loss due to high directivity and sensitivity to blockages, limiting its adoption within small-scale deployments. To enhance the coverage of mmWave communication in 5G and beyond, it is promising to deploy a large number of reconfigurable intelligent surfaces (RISs) that passively reflect mmWave signals towards desired directions. With this motivation, in this work we study the coverage of an RIS-assisted large-scale mmWave cellular network using stochastic geometry, and derive the peak reflection power expression of an RIS and the downlink signal-to-interference ratio (SIR) coverage expression in closed forms. These analytic results clarify the effectiveness of deploying RISs in the mmWave SIR coverage enhancement, while unveiling the major role of the density ratio between active base stations (BSs) and passive RISs. Furthermore, the results show that deploying passive reflectors is as effective as equipping BSs with more active antennas in the mmWave coverage enhancement. Simulation results confirm the tightness of the closed form expressions, corroborating our major findings based on the derived expressions.

eess.SP↗

arXiv2021

Mean-Field Game-Theoretic Edge Caching

In this book chapter, we study a problem of distributed content caching in an ultra-dense edge caching network (UDCN), in which a large number of small base stations (SBSs) prefetch popular files to cope with the ever-growing user demand in 5G and beyond. In a UDCN, even a small misprediction of user demand may render a large amount of prefetched data obsolete. Furtherproacmore, the interference variance is high due to the short inter-SBS distances, making it difficult to quantify data downloading rates. Lastly, since the caching decision of each SBS interacts with those of all other SBSs, the problem complexity of exponentially increases with the number of SBSs, which is unfit for UDCNs. To resolve such challenging issues while reflecting time-varying and location-dependent user demand, we leverage mean-field game (MFG) theory through which each SBS interacts only with a single virtual SBS whose state is drawn from the state distribution of the entire SBS population, i.e., mean-field (MF) distribution. This MF approximation asymptotically guarantees achieving the epsilon Nash equilibrium as the number of SBSs approaches infinity. To describe such an MFG-theoretic caching framework, this chapter aims to provide a brief review of MFG, and demonstrate its effectiveness for UDCNs.

cs.IT↗