Graph-Based Feature Weight Optimisation and Classification of Continuous Seismic Sensor Array Recordings.

Slope instabilities caused by heavy rainfall, man-made activity or earthquakes can be characterised by seismic events. To minimise mortality and infrastructure damage, a good understanding of seismic signal properties characterising slope failures is therefore crucial to classify seismic events recorded from continuous recordings effectively. However, there are limited contributions towards understanding the importance of feature selection for the classification of seismic signals from continuous noisy recordings from multiple channels/sensors. This paper first proposes a novel multi-channel event-detection scheme based on Neyman-Pearson lemma and Multi-channel Coherency Migration (MCM) on the stacked signal across multi-channels. Furthermore, this paper adapts graph-based feature weight optimisation as feature selection, exploiting the signal's physical characteristics, to improve signal classification. Specifically, we alternatively optimise the feature weight and classification label with graph smoothness and semidefinite programming (SDP). Experimental results show that with expert interpretation, compared with the conventional short-time average/long-time average (STA/LTA) detection approach, our detection method identified 614 more seismic events in five days. Furthermore, feature selection, especially via graph-based feature weight optimisation, provides more focused feature sets with less than half of the original number of features, at the same time enhancing the classification performance; for example, with feature selection, the Graph Laplacian Regularisation classifier (GLR) raised the rockfall and slide quake sensitivities to 92% and 88% from 89% and 85%, respectively.

A Gradient-Based Method for Robust SensorSelection in Hypothesis Testing.

This paper considers the binary Gaussian distribution robust hypothesis testing under aBayesian optimal criterion in the wireless sensor network (WSN). The distribution covariance matrixunder each hypothesis is known, while the distribution mean vector under each hypothesis driftsin an ellipsoidal uncertainty set. Because of the limited bandwidth and energy, we aim at seeking asubset of p out of m sensors such that the best detection performance is achieved. In this setup, theminimax robust sensor selection problem is proposed to deal with the uncertainties of distributionmeans. Following a popular method, minimizing the maximum overall error probability with respectto the selection matrix can be approximated by maximizing the minimum Chernoff distance betweenthe distributions of the selected measurements under null hypothesis and alternative hypothesis tobe detected. Then, we utilize Danskin's theorem to compute the gradient of the objective functionof the converted maximization problem, and apply the orthogonal constraint-preserving gradientalgorithm (OCPGA) to solve the relaxed maximization problem without 0/1 constraints. It is shownthat the OCPGA can obtain a stationary point of the relaxed problem. Meanwhile, we provide thecomputational complexity of the OCPGA, which is much lower than that of the existing greedyalgorithm. Finally, numerical simulations illustrate that, after the same projection and refinementphases, the OCPGA-based method can obtain better solutions than the greedy algorithm-basedmethod but with up to 48.72% shorter runtimes. Particularly, for small-scale problems, the OCPGA-based method is able to attain the globally optimal solution.

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.

Anchor-Free Correlated Topic Modeling.

In topic modeling, identifiability of the topics is an essential issue. Many topic modeling approaches have been developed under the premise that each topic has a characteristic anchor word that only appears in that topic. The anchor-word assumption is fragile in practice, because words and terms have multiple uses; yet it is commonly adopted because it enables identifiability guarantees. Remedies in the literature include using three- or higher-order word co-occurence statistics to come up with tensor factorization models, but such statistics need many more samples to obtain reliable estimates, and identifiability still hinges on additional assumptions, such as consecutive words being persistently drawn from the same topic. In this work, we propose a new topic identification criterion using second order statistics of the words. The criterion is theoretically guaranteed to identify the underlying topics even when the anchor-word assumption is grossly violated. An algorithm based on alternating optimization, and an efficient primal-dual algorithm are proposed to handle the resulting identification problem. The former exhibits high performance and is completely parameter-free; the latter affords up to 200 times speedup relative to the former, but requires step-size tuning and a slight sacrifice in accuracy. A variety of real text copora are employed to showcase the effectiveness of the approach, where the proposed anchor-free method demonstrates substantial improvements compared to a number of anchor-word based approaches under various evaluation metrics.