查看更多>>摘要：Newton-type greedy pursuit methods have been shown to work favorably for cardinality-constrained sparse learning problems. The appealing sparsity recovery performance of the existing Newton-type greedy pursuit methods, however, is typically guaranteed within a local neighborhood around the target solution. To address this limitation, we present in this paper a novel approximate Newton pursuit method for sparse learning with linear models. The computation procedure of our method iterates between constructing an inexact Newton-type quadratic majorization to the global empirical risk and solving the quadratic approximation via iterative hard thresholding. Provable global guarantees on mean squared prediction error, which is less understood for prior methods, are provided for our method. Numerical evidence is provided to show the advantages of our approach over the prior methods. (c) 2022 Published by Elsevier Ltd.
查看更多>>摘要：Dimensionality reduction plays a crucial role in classification, object detection, and pattern recognition tasks. Its main objective is to decrease the dimension of the original data while retaining the most distinctive information. With the emergence of deep learning, an autoencoder has become a state-of-the-art non-linear dimensionality-reduction method. Nonetheless, as the existing autoencoder models are devised to follow the data distribution and employ similarity techniques, preserving distinctive information can be problematic. To tackle this issue, we propose super-encoder (SE) networks trained in a supervised and cooperative manner. The SE consists of an encoder, separator, and decoder networks. The encoder combined with separator networks are dedicated to generating separable latent representation based on the label, and the decoder network should be able to reconstruct it to the original data simultaneously. Herein, we introduce a novel cooperative learning mechanism with a new loss function; therefore, the encoder, separator, and decoder networks can cooperate to achieve these objectives. Extensive experiments using benchmark datasets were conducted. The results indicated that the SE is more effective in extracting separable latent code than the existing supervised and unsupervised dimensionality-reduction models. Furthermore, as a generator, it can obtain highly competitive realistic images. (c) 2022 Elsevier Ltd. All rights reserved.
查看更多>>摘要：Large volumes of training data introduce high computational cost in instance-based classification. Data reduction algorithms select or generate a small (condensing) set of representative training prototypes from the available training data. The Reduction by Space Partitioning algorithm is one of the most wellknown prototype generation algorithms that repetitively divides the original training data into subsets. This partitioning process needs to identify the diameter of each subset, i.e., its two farthest instances. This is a costly process since it requires the calculation of all distances between the instances in each subset. The paper introduces two new very fast variations that, instead of computing the actual diameter of a subset, choose a pair of distant-enough instances. The first variation uses instances belonging to an exact 3d convex hull of the subset, while the second one uses instances belonging to the minimum bounding rectangle of the subset. Our experimental study shows that the new variations vastly outperform the original algorithm without a penalty in classification accuracy and reduction rate. 0 2022 Elsevier Ltd. All rights reserved.
Sun, WeiGong, DongShi, Javen Qinfengvan den Hengel, Anton...
14页
查看更多>>摘要：Video super-resolution aims to recover the high-resolution (HR) contents from the low-resolution (LR) observations relying on compositing the spatial-temporal information in the LR frames. It is crucial to model the spatial-temporal information jointly since the video sequences are three-dimensional spatial temporal signals. Compared with explicitly estimating motions between the 2D frames, 3D convolutional neural networks (CNNs) have been shown its efficiency and effectiveness for video super-resolution (SR), as a natural way of spatial-temporal data modelling. Though promising, the performance of 3D CNNs is still far from satisfactory. The high computational and memory requirements limit the development of more advanced designs to extract and fuse the information from a larger spatial and temporal scale. We thus propose a Mixed Spatial-Temporal Convolution (MSTC) block that simultaneously extracts the spatial information and the supplemented temporal dependency among frames by jointly applying 2D and 3D convolution. To further fuse the learned features corresponding to different frames, we propose a novel similarity-based selective features strategy, unlike precious methods directly stacking the learned features. Additionally, an attention-based motion compensation module is applied to alleviate the influence of misalignment between frames. Experiments on three widely used benchmark datasets and real-world dataset show that, relying on superior feature extraction and fusion ability, the proposed network can outperform previous state-of-the-art methods, especially for recovering the confusing details. (c) 2022 Elsevier Ltd. All rights reserved.
Wang, HaoLin, GuoshengHoi, Steven C. H.Miao, Chunyan...
9页
查看更多>>摘要：Recipe generation from food images and ingredients is a challenging task, which requires the interpretation of the information from another modality. Different from the image captioning task, where the captions usually have one sentence, cooking instructions contain multiple sentences and have obvious structures. To help the model capture the recipe structure and avoid missing some cooking details, we propose a novel framework: Decomposing Generation Networks (DGN) with structure prediction, to get more structured and complete recipe generation outputs. Specifically, we split each cooking instruction into several phases, and assign different sub-generators to each phase. Our approach includes two novel ideas: (i) learning the recipe structures with the global structure prediction component and (ii) producing recipe phases in the sub-generator output component based on the predicted structure. Extensive experiments on the challenging large-scale Recipe1M dataset validate the effectiveness of our proposed model, which improves the performance over the state-of-the-art results. (c) 2022 Elsevier Ltd. All rights reserved.
查看更多>>摘要：Recently, structured proximity matrix learning, which aims to learn a structured proximity matrix with explicit clustering structures from the first-order proximity matrix, has become the mainstream of graph-based clustering. However, the first-order proximity matrix always lacks several must-links compared to the groundtruth in real-world data, which results in a mismatched problem and affects the clustering performance. To alleviate this problem, this work introduces the high-order proximity to structured prox-imity matrix learning, and explores a novel framework named Adaptive-Order Proximity Learning (AOPL) to learn a consensus structured proximity matrix from the proximities of multiple orders. To be specific, AOPL selects the appropriate orders first, then assigns weights to these selected orders adaptively. In this way, a consensus structured proximity matrix is learned from the proximity matrices of appropriate orders. Based on AOPL framework, two practical models with different properties are derived, namely AOPL-Root and AOPL-Log. Besides, AOPL and the derived models are regarded as the same optimization problem subjected to some slightly different constraints. An efficient algorithm is proposed to solve them and the corresponding theoretical analyses are provided. Extensive experiments on several real-world datasets demonstrate superb performance of our model. (c) 2022 Elsevier Ltd. All rights reserved.
查看更多>>摘要：This paper describes probability forecasting systems that are universal , or universally consistent , in the sense of being consistent under any data-generating distribution, assuming that the observations are produced independently in the IID fashion. The notion of universal consistency is asymptotic and does not imply any small-sample guarantees of validity. On the other hand, the method of conformal prediction has been recently adapted to producing predictive distributions that satisfy a natural property of small sample validity, namely they are automatically probabilistically calibrated. The main result of the paper is the existence of universal conformal predictive systems, which output predictive distributions that are both probabilistically calibrated and universally consistent. (c) 2022 Elsevier Ltd. All rights reserved.
Du, ShuaiyuanHong, ChaoyiChen, YinpengCao, Zhiguo...
12页
查看更多>>摘要：In this paper, we address the problem of novelty detection whose goal is to recognize instances from unseen classes during testing. Our key idea is to leverage the inconsistency between class similarity and (latent) attribute similarity. We are motivated by the observation that a novel class may holistically appear like a certain known class (class-level reference) but often exhibits unique properties similar to others (attribute-level references). That is, the related class-and attribute-level references are often inconsistent for a novel class. A new two-stage Class-Attribute Inconsistency Learning network (CAILNet) is proposed to explore class-attribute inconsistency for novelty detection. Stage one aims to learn both class and attribute features based on the class labels and fake attribute labels, and stage two aims to search for the corresponding references and make fine-grained comparisons for final novelty decision. Empirically we conduct comprehensive experiments on three benchmark datasets, and demonstrate state-of-the-art performance. (c) 2022 Elsevier Ltd. All rights reserved.
NSTL
Elsevier