We consider a state-space modeling approach for online estimation of a signal's instantaneous frequency (IF). We take into account the general case wherein the IF may vary over time irregularly and, therefore, given several plausible models, there is uncertainty which is the best model to use. We address this dynamical model uncertainty problem using a strategy called dynamic model averaging (DMA), in which a set of state-space models is combined with a Markov chain model for the correct model. The model transition process is specified in terms of forgetting, leading to a highly parsimonious representation. We provide six candidate models, each representing a specific evolution law of the IF. A novel particle filtering algorithm is proposed to implement this DMA strategy. Simulation results show that our method can yield accurate IF estimates in challenging settings, where the IF structure embedded in the signal changes abruptly and irregularly in a low signal-noise-ratio environment.

We are concerned with tracking an object of interest in a video stream. We propose an algorithm that is robust against occlusion, the presence of confusing colors, abrupt changes in the object features and changes in scale. We develop the algorithm within a Bayesian modeling framework. The state-space model is used for capturing the temporal correlation in the sequence of frame images by modeling the underlying dynamics of the tracking system. The Bayesian model averaging (BMA) strategy is proposed for fusing multiclue information in the observations. Any number of object features is allowed to be involved in the proposed framework. Every feature represents one source of information to be fused and is associated with an observation model. The state inference is performed by employing the particle filter methods. In comparison with the related approaches, the BMA-based tracker is shown to have robustness, expressivity, and comprehensibility.

Averaging neural network weights sampled by a backbone stochastic gradient descent (SGD) is a simple-yet-effective approach to assist the backbone SGD in finding better optima, in terms of generalization. From a statistical perspective, weight-averaging contributes to variance reduction. Recently, a well-established stochastic weight-averaging (SWA) method was proposed, which featured the application of a cyclical or high-constant (CHC) learning-rate schedule for generating weight samples for weight-averaging. Then, a new insight on weight-averaging was introduced, which stated that weight average assisted in discovering a wider optima and resulted in better generalization. We conducted extensive experimental studies concerning SWA, involving 12 modern deep neural network model architectures and 12 open-source image, graph, and text datasets as benchmarks. We disentangled the contributions of the weight-averaging operation and the CHC learning-rate schedule for SWA, showing that the weight-averaging operation in SWA still contributed to variance reduction, and the CHC learning-rate schedule assisted in exploring the parameter space more widely than the backbone SGD, which could be be under-fitted due to a lack of training budget. We then presented an algorithm termed periodic SWA (PSWA) that comprised a series of weight-averaging operations to exploit such wide parameter space structures as explored by the CHC learning-rate schedule, and we empirically demonstrated that PSWA outperformed its backbone SGD remarkably.

This paper discusses the relationship between data science and population-based algorithms, which include swarm intelligence and evolutionary algorithms. We reviewed two categories of literature, which include population-based algorithms solving data analysis problem and utilizing data analysis methods in population-based algorithms. With the exponential increment of data, the data science, or more specifically, the big data analytics has gained increasing attention among researchers. New and more efficient algorithms should be designed to handle this massive data problem. Based on the combination of population-based algorithms and data mining techniques, we understand better the insights of data analytics, and design more efficient algorithms to solve real-world big data analytics problems. Also, the weakness and strength of population-based algorithms could be analyzed via the data analytics along the optimization process, a crucial entity in population-based algorithms.