Domain generalization (DG), i.e., out-of-distribution generalization, has attracted increased interests in recent years. Domain generalization deals with a challenging setting where one or several different but related domain(s) are given, and the goal is to learn a model that can generalize to an unseen test domain. For years, great progress has been achieved. This paper presents the first review for recent advances in domain generalization. First, we provide a formal definition of domain generalization and discuss several related fields. Then, we categorize recent algorithms into three classes and present them in detail: data manipulation, representation learning, and learning strategy, each of which contains several popular algorithms. Third, we introduce the commonly used datasets and applications. Finally, we summarize existing literature and present some potential research topics for the future.

No abstract

In recent years, the delay in sowing has become a major obstacle to high wheat yield in Jiangsu Province, one of the major wheat producing areas in China; hence, it is necessary to screen wheat varieties are resilient for late sowing. This study aimed to provide an effective, fast, and non-destructive monitoring method of soil plant analysis development (SPAD) values, which can represent leaf chlorophyll contents, for late-sown winter wheat variety screening. This study acquired multispectral images using an unmanned aerial vehicle (UAV) at the overwintering stage of winter wheat growth, and further processed these images to extract reflectance of five single spectral bands and calculated 26 spectral vegetation indices. Based on these 31 variables, this study combined three variable selection methods (i.e., recursive feature elimination (RFE), random forest (RF), and Pearson correlation coefficient (r)) with four machine learning algorithms (i.e., random forest regression (RFR), linear kernel-based support vector regression (SVR), radial basis function (RBF) kernel-based SVR, and sigmoid kernel-based SVR), resulted in seven SVR models (i.e., RFE-SVR_linear, RF-SVR_linear, RF-SVR_RBF, RF-SVR_sigmoid, r-SVR_linear, r-SVR_RBF, and r-SVR_sigmoid) and three RFR models (i.e., RFE-RFR, RF-RFR, and r-RFR). The performances of the 10 machine learning models were evaluated and compared with each other according to the achieved coefficient of determination (R2), residual prediction deviation (RPD), root mean square error (RMSE), and relative RMSE (RRMSE) in SPAD estimation. Of the 10 models, the best one was the RF-SVR_sigmoid model, which was the combination of the RF variable selection method and the sigmoid kernel-based SVR algorithm. It achieved high accuracy in estimating SPAD values of the wheat canopy (R2 = 0.754, RPD = 2.017, RMSE = 1.716 and RRMSE = 4.504%). The newly developed UAV- and machine learning-based model provided a promising and real time method to monitor chlorophyll contents at the overwintering stage, which can benefit late-sown winter wheat variety screening.

This paper proposes the use of Recurrent Neural Networks (RNNs) with Long Short-Term Memory (LSTM) units for determining whether Mandarin-speaking individuals are afflicted with a form of Dysarthria based on samples of syllable pronunciations. Several LSTM network architectures are evaluated on this binary classification task, using accuracy and Receiver Operating Characteristic (ROC) curves as metrics. The LSTM models are shown to significantly improve upon a baseline fully connected network, reaching over 90% area under the ROC curve on the task of classifying new speakers, when a sufficient number of cepstrum coefficients are used. The results show that the LSTM's ability to leverage temporal information within its input makes for an effective step in the pursuit of accessible Dysarthria diagnoses.

No abstract