Search

Differentiable architecture search (DARTS) transforms architectural optimization into a super network optimization by stacking two cells (2 c.). However, repeatedly stacking two cells is a suboptimal operation since cells in different depths should be various. Besides, we find that the performance is slightly improved by increasing the number of searched cells (e.g., from 2 c. to 5 c.), but it leads to uneven resource allocation. This letter proposes a regularized DARTS (RDARTS) to adjust the architectural differences and balance degrees of freedom and resource allocation. Specifically, we use separate architectural parameters for two reduction cells and three normal cells, and then propose an Reg distance to calculate the difference between cells. We design a new validation loss which is the weighting of cross-entropy and Reg loss and introduce an adaptive adjustment method. Results show that RDARTS achieves the top-1 accuracy of 97.64% and 75.8% on CIFAR and ImageNet. [ABSTRACT FROM AUTHOR]

• We enlarge search space by introducing the Cartesian product of operators. • We propose a factorized search to search separately in two subspaces. • We optimize architectural parameters sequentially, avoiding undesirable competition. • We perform diagnostic experiments and explanations, demonstrating our superiority. • Our approach achieves SOTA performance on CIFAR10, ImageNet, and COCO. Neural Architecture Search (NAS) replaces manually designed networks with automatically searched networks and has become a hot topic in machine learning and computer vision. One key factor of NAS, scaling up the search space by adding operators, could help bring about more possibilities for effective architectures. However, existing works require high search costs and are prone to make confused selections caused by competition among similar operators. This paper presents a scalable NAS that utilizes a factorized method, which improves existing art in the following aspects. (1) Our work explores a broader search space. We construct an ample search space through the cartesian product of activation operators and regular operators. (2) Our work experiences a limited computation burden even though searching in a more extensive search space. We factorize a search space into two subspaces and adopt different architectural parameters to control corresponding subspaces. (3) Our work avoids competition among similarly combined operators (e.g. , ReLU & sep-conv-3x3 and ReLU6 & sep-conv-3x3 and PReLU & sep-conv-3x3). That is because our architectural parameters are optimized sequentially. Experimental results show that our approach achieves SOTA on CIFAR10 (2. 38 % test error) and ImageNet (23. 8 % test error). Furthermore, the excellent performance of 35. 8 % (AP) and 55. 7 % ( AP 50 ) on COCO further proves the superiority of our factorized method. [ABSTRACT FROM AUTHOR]

Statistical-learning-based object detection is an important topic in computer vision. It learns visual representation from annotated exemplars to identify semantic defined objects in images. High-performance object detection is usually carried out in feature space and effective feature representation can improve the performance significantly. Feature representation is the encoding process which maps raw image pixels inside local regions into discriminant feature space. The motivation of this paper is to present a review on feature representation in recent object detection methods. Visual features applied in object detection are categorized according to the differences in computation and visual properties. The most valued features are introduced and discussed in detail. Representative extensions are introduced briefly for comparison. Descriptive power, robustness, compactness as well as computational efficiency are viewed as important properties. According to these properties, discussions are presented on the advantages and drawbacks of features. Besides, generic techniques such as dimension reduction and combination are introduced. Through this review, we would like to draw the feature sketch and provide new insights for feature utilization, in order to tackle future challenges of object detection. [ABSTRACT FROM AUTHOR]

Feature detection is a fundamental and important problem in computer vision and image processing. It is a low-level processing step which serves as the essential part for computer vision based applications. The goal of this paper is to present a survey of recent progress and advances in visual feature detection. Firstly we describe the relations among edges, corners and blobs from the psychological view. Secondly we classify the algorithms in detecting edges, corners and blobs into different categories and provide detailed descriptions for representative recent algorithms in each category. Considering that machine learning becomes more involved in visual feature detection, we put more emphasis on machine learning based feature detection methods. Thirdly, evaluation standards and databases are also introduced. Through this survey we would like to present the recent progress in visual feature detection and identify future trends as well as challenges. [ABSTRACT FROM AUTHOR]

• Multi-view Adaptive Semi-supervised Feature Selection framework is proposed. • The self-paced learning is applied to semi-supervised feature selection. • Multi-view learning is utilized to enhance the feature selection performance. • An iterative algorithm is proposed with convergence and complexity analysis. • Experiments show MASFS can obtain good semi-supervised feature selection performance. In recent decades, semi-supervised feature selection (SSFS) has gained extensive research in the field of machine learning and computer vision. Most SSFS algorithms are based on graph-based semi-supervised learning (GSSL), and their performance depend heavily on the quality of the Laplacian weight graph. However, the Laplacian weight graph can't be changed once it is constructed, which greatly restricts the performance of SSFS. To address this defect, in this paper we propose a novel Multi-view Adaptive Semi-supervised Feature Selection (MASFS) algorithm, which introduces the self-paced learning (SPL) into SSFS to make the Laplacian weight graph adaptively change according to the current predicted information. Meanwhile, the MASFS algorithm utilizes the multi-view learning to effectively explore the complementary and related information contained in different views to enhance SSFS performance. We propose a valid iterative algorithm for optimizing the objective function, followed by the convergence analysis and the complexity analysis. To illustrate the effectiveness of the MASFS algorithm, some experiments are carried out on NUS-WIDE dataset and MSRA-MM2.0 dataset and the experimental results indicate that MASFS has better performance than other SSFS algorithms based on GSSL. [ABSTRACT FROM AUTHOR]