In federated learning (FL), it is generally 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, ML tasks are expected to proceed in an online fashion, wherein data samples are generated as a function of time and each client has to predict a label (or make a decision) upon receiving an incoming data. To this end, online FL (OFL) has been introduced, which aims at learning a sequence of global models from distributed streaming data such that a cumulative regret is minimized. In this framework, the vanilla method (named FedOGD) by combining online gradient descent and model averaging, which is regarded as the counterpart of FedSGD in the standard FL. Despite its asymptotic optimality, FedOGD suffers from high communication costs. In this paper, we present a communication-efficient OFL method by means of intermittent transmission (enabled by client subsampling and periodic transmission) and gradient quantization. For the first time, we derive the regret bound which can reflect the impact of data-heterogeneity and communication-efficient techniques. Based on our tighter analysis, we optimize the key parameters of OFedIQ such as sampling rate, transmission period, and quantization bits. Also, we prove that the optimized OFedIQ asymptotically achieves the performance of FedOGD while reducing the communication costs by 99%. Via experiments with real datasets, we validate the effectiveness of our algorithm on various online ML tasks.

We investigate an online federated learning (in short, OFL), in which many edge nodes receive their own time-series data and train a sequence of global models under the orchestration of a central server while keeping data localized. In this framework, we propose a communication efficient federated quantized online gradient descent (FedQOGD) by means of a stochastic quantization and partial node participation. We theoretically prove that FedQOGD over T time slots can achieve an optimal sublinear regret bound ${\mathcal{O}}(\sqrt T )$ for any quantization level (e.g., 1-level quantization), even when every node can participate in a learning process sporadically. Our analysis reveals that FedQOGD yields the same asymptotic performance as the centralized counterpart (i.e., all local data are gathered at the central server) while having a low-communication overhead and preserving an edge-node privacy. Finally, we verify the effectiveness of our algorithm via experiments with a real-world MNIST dataset on online classification task.

Providing appropriate help is important in smartphone development as smartphones have become increasingly complex owing to their large number of features. To determine the appropriate help content, numerous studies on contextual help systems have been conducted; however, few studies have been concerned with user manual content. Thus, to provide effective user manuals, we focused on content prioritization, considering the usage pattern. Specifically, we calculated the vector representation of each element of the usage pattern and adopted a heterogeneous embedding approach. Moreover, we embedded the entire usage pattern using RNN-SVAE to calculate a user modeling value for representing user interests. Additionally, we trained InfoGAN (a generative adversarial network) to predict the usage of the user manual, and we prioritized and re-organized its content accordingly. Experiments demonstrated that, compared with existing benchmark methods, the proposed method can achieve better content-usage prediction and more effective prioritization of the top-k contents.

In this paper, a gene-handling method for evolutionary algorithms (EAs) is proposed. Such algorithms are characterized by a nonanalytic optimization process when dealing with complex systems as multiple behavioral responses occur in the realization of intelligent tasks. In generic EAs which optimize internal parameters of a given system, evaluation and selection are performed at the chromosome level. When a survived chromosome includes noneffective genes, the solution can be trapped in a local optimum during evolution, which causes an increase in the uncertainty of the results and reduces the quality of the overall system. This phenomenon also results in an unbalanced performance of partial behaviors. To alleviate this problem, a score-based resampling method is proposed, where a score function of a gene is introduced as a criterion of handling genes in each allele. The proposed method was empirically evaluated with various test functions, and the results show its effectiveness.