Song‐Nam Hong

Dohyeok Kwon, Jonghwan Park, Songnam Hong

In federated learning (FL), it is commonly assumed that all data are placed at clients in the beginning of machine learning (ML) optimization (i.e., offline learning). However, in many real-world applications, it is expected to proceed in an online fashion. To this end, online FL (OFL) has been introduced, which aims at learning a sequence of global models from decentralized streaming data such that the so-called cumulative regret is minimized. Combining online gradient descent and model averaging, in this framework, FedOGD is constructed as the counterpart of FedSGD in FL. While it can enjoy an optimal sublinear regret, FedOGD suffers from heavy communication costs. In this paper, we present a communication-efficient method (named OFedIQ) by means of intermittent transmission (enabled by client subsampling and periodic transmission) and quantization. For the first time, we derive the regret bound that captures the impact of data-heterogeneity and the communication-efficient techniques. Through this, we efficiently optimize the parameters of OFedIQ such as sampling rate, transmission period, and quantization levels. Also, it is proved that the optimized OFedIQ can asymptotically achieve the performance of FedOGD while reducing the communication costs by 99%. Via experiments with real datasets, we demonstrate the effectiveness of the optimized OFedIQ.

Ukjo Hwang, Songnam Hong

We investigate a Federated Reinforcement Learning with Environment Heterogeneity (FRL-EH) framework, where local environments exhibit statistical heterogeneity. Within this framework, agents collaboratively learn a global policy by aggregating their collective experiences while preserving the privacy of their local trajectories. To better reflect real-world scenarios, we introduce a robust FRL-EH framework by presenting a novel global objective function. This function is specifically designed to optimize a global policy that ensures robust performance across heterogeneous local environments and their plausible perturbations. We propose a tabular FRL algorithm named FedRQ and theoretically prove its asymptotic convergence to an optimal policy for the global objective function. Furthermore, we extend FedRQ to environments with continuous state space through the use of expectile loss, addressing the key challenge of minimizing a value function over a continuous subset of the state space. This advancement facilitates the seamless integration of the principles of FedRQ with various Deep Neural Network (DNN)-based RL algorithms. Extensive empirical evaluations validate the effectiveness and robustness of our FRL algorithms across diverse heterogeneous environments, consistently achieving superior performance over the existing state-of-the-art FRL algorithms.

Ukjo Hwang, Songnam Hong

Overestimation is a fundamental characteristic of model-free reinforcement learning (MF-RL), arising from the principles of temporal difference learning and the approximation of the Q-function. To address this challenge, we propose a novel moderate target in the Q-function update, formulated as a convex optimization of an overestimated Q-function and its lower bound. Our primary contribution lies in the efficient estimation of this lower bound through the lower expectile of the Q-value distribution conditioned on a state. Notably, our moderate target integrates seamlessly into state-of-the-art (SOTA) MF-RL algorithms, including Deep Deterministic Policy Gradient (DDPG) and Soft Actor Critic (SAC). Experimental results validate the effectiveness of our moderate target in mitigating overestimation bias in DDPG, SAC, and distributional RL algorithms.

Ukjo Hwang, Songnam Hong

Robust reinforcement learning (RRL) aims at seeking a robust policy to optimize the worst case performance over an uncertainty set of Markov decision processes (MDPs). This set contains some perturbed MDPs from a nominal MDP (N-MDP) that generate samples for training, which reflects some potential mismatches between training (i.e., N-MDP) and true environments. In this paper we present an elaborated uncertainty set by excluding some implausible MDPs from the existing sets. Under this uncertainty set, we develop a sample-based RRL algorithm (named ARQ-Learning) for tabular setting and characterize its finite-time error bound. Also, it is proved that ARQ-Learning converges as fast as the standard Q-Learning and robust Q-Learning while ensuring better robustness. We introduce an additional pessimistic agent which can tackle the major bottleneck for the extension of ARQ-Learning into the cases with larger or continuous state spaces. Incorporating this idea into RL algorithms, we propose double-agent algorithms for model-free RRL. Via experiments, we demonstrate the effectiveness of the proposed algorithms.

Jeongmin Chae, Songnam Hong

We consider the problem of learning a nonlinear function over a network of learners in a fully decentralized fashion. Online learning is additionally assumed, where every learner receives continuous streaming data locally. This learning model is called a fully distributed online learning (or a fully decentralized online federated learning). For this model, we propose a novel learning framework with multiple kernels, which is named DOMKL. The proposed DOMKL is devised by harnessing the principles of an online alternating direction method of multipliers and a distributed Hedge algorithm. We theoretically prove that DOMKL over T time slots can achieve an optimal sublinear regret, implying that every learner in the network can learn a common function which has a diminishing gap from the best function in hindsight. Our analysis also reveals that DOMKL yields the same asymptotic performance of the state-of-the-art centralized approach while keeping local data at edge learners. Via numerical tests with real datasets, we demonstrate the effectiveness of the proposed DOMKL on various online regression and time-series prediction tasks.

Jeongmin Chae, Songnam Hong

Online federated learning (OFL) becomes an emerging learning framework, in which edge nodes perform online learning with continuous streaming local data and a server constructs a global model from the aggregated local models. Online multiple kernel learning (OMKL), using a preselected set of P kernels, can be a good candidate for OFL framework as it has provided an outstanding performance with a low-complexity and scalability. Yet, an naive extension of OMKL into OFL framework suffers from a heavy communication overhead that grows linearly with P. In this paper, we propose a novel multiple kernel-based OFL (MK-OFL) as a non-trivial extension of OMKL, which yields the same performance of the naive extension with 1/P communication overhead reduction. We theoretically prove that MK-OFL achieves the optimal sublinear regret bound when compared with the best function in hindsight. Finally, we provide the numerical tests of our approach on real-world datasets, which suggests its practicality.

Seunghoon Lee, Chanho Park, Song-Nam Hong et al.

Federated learning is a privacy-preserving and distributed training method using heterogeneous data sets stored at local devices. Federated learning over wireless networks requires aggregating locally computed gradients at a server where the mobile devices send statistically distinct gradient information over heterogenous communication links. This paper proposes a Bayesian federated learning (BFL) algorithm to aggregate the heterogeneous quantized gradient information optimally in the sense of minimizing the mean-squared error (MSE). The idea of BFL is to aggregate the one-bit quantized local gradients at the server by jointly exploiting i) the prior distributions of the local gradients, ii) the gradient quantizer function, and iii) channel distributions. Implementing BFL requires high communication and computational costs as the number of mobile devices increases. To address this challenge, we also present an efficient modified BFL algorithm called scalable-BFL (SBFL). In SBFL, we assume a simplified distribution on the local gradient. Each mobile device sends its one-bit quantized local gradient together with two scalar parameters representing this distribution. The server then aggregates the noisy and faded quantized gradients to minimize the MSE. We provide a convergence analysis of SBFL for a class of non-convex loss functions. Our analysis elucidates how the parameters of communication channels and the gradient priors affect convergence. From simulations, we demonstrate that SBFL considerably outperforms the conventional sign stochastic gradient descent algorithm when training and testing neural networks using MNIST data sets over heterogeneous wireless networks.

Jeongmin Chae, Songnam Hong

In the Internet-of-Things (IoT) systems, there are plenty of informative data provided by a massive number of IoT devices (e.g., sensors). Learning a function from such data is of great interest in machine learning tasks for IoT systems. Focusing on streaming (or sequential) data, we present a privacy-preserving distributed online learning framework with multiplekernels (named DOMKL). The proposed DOMKL is devised by leveraging the principles of an online alternating direction of multipliers (OADMM) and a distributed Hedge algorithm. We theoretically prove that DOMKL over T time slots can achieve an optimal sublinear regret, implying that every learned function achieves the performance of the best function in hindsight as in the state-of-the-art centralized online learning method. Moreover, it is ensured that the learned functions of any two neighboring learners have a negligible difference as T grows, i.e., the so-called consensus constraints hold. Via experimental tests with various real datasets, we verify the effectiveness of the proposed DOMKL on regression and time-series prediction tasks.

Jeongmin Chae, Songnam Hong

We study a pool-based sequential active learning (AL), in which one sample is queried at each time from a large pool of unlabeled data according to a selection criterion. For this framework, we propose two selection criteria, named expected-kernel-discrepancy (EKD) and expected-kernel-loss (EKL), by leveraging the particular structure of multiple kernel learning (MKL). Also, it is identified that the proposed EKD and EKL successfully generalize the concepts of popular query-by-committee (QBC) and expected-model-change (EMC), respectively. Via experimental results with real-data sets, we verify the effectiveness of the proposed criteria compared with the existing methods.

Songnam Hong, Jeongmin Chae

Online multiple kernel learning (OMKL) has provided an attractive performance in nonlinear function learning tasks. Leveraging a random feature approximation, the major drawback of OMKL, known as the curse of dimensionality, has been recently alleviated. In this paper, we introduce a new research problem, termed (stream-based) active multiple kernel learning (AMKL), in which a learner is allowed to label selected data from an oracle according to a selection criterion. This is necessary in many real-world applications as acquiring true labels is costly or time-consuming. We prove that AMKL achieves an optimal sublinear regret, implying that the proposed selection criterion indeed avoids unuseful label-requests. Furthermore, we propose AMKL with an adaptive kernel selection (AMKL-AKS) in which irrelevant kernels can be excluded from a kernel dictionary 'on the fly'. This approach can improve the efficiency of active learning as well as the accuracy of a function approximation. Via numerical tests with various real datasets, it is demonstrated that AMKL-AKS yields a similar or better performance than the best-known OMKL, with a smaller number of labeled data.