Search

The human visual system can recognize object categories accurately and efficiently and is robust to complex textures and noises. To mimic the analogy-detail dual-pathway human visual cognitive mechanism revealed in recent cognitive science studies, in this article, we propose a novel convolutional neural network (CNN) architecture named analogy-detail networks (ADNets) for accurate object recognition. ADNets disentangle the visual information and process them separately using two pathways: the analogy pathway extracts coarse and global features representing the gist (i.e., shape and topology) of the object, while the detail pathway extracts fine and local features representing the details (i.e., texture and edges) for determining object categories. We modularize the architecture and encapsulate the two pathways into the analogy-detail block as the CNN building block to construct ADNets. For implementation, we propose a general principle that transmutes typical CNN structures into the ADNet architecture and applies the transmutation on representative baseline CNNs. Extensive experiments on CIFAR10, CIFAR100, street view house numbers, and ImageNet data sets demonstrate that ADNets significantly reduce the test error rates of the baseline CNNs by up to 5.76% and outperform other state-of-the-art architectures. Comprehensive analysis and visualizations further demonstrate that ADNets are interpretable and have a better shape-texture tradeoff for recognizing the objects with complex textures.

In this paper, we propose a novel single-task continual learning framework named Bi-Objective Continual Learning (BOCL). BOCL aims at both consolidating historical knowledge and learning from new data. On one hand, we propose to preserve the old knowledge using a small set of pillars, and develop the pillar consolidation (PLC) loss to preserve the old knowledge and to alleviate the catastrophic forgetting problem. On the other hand, we develop the contrastive pillar (CPL) loss term to improve the classification performance, and examine several data sampling strategies for efficient onsite learning from ‘new’ with a reasonable amount of computational resources. Comprehensive experiments on CIFAR10/100, CORe50 and a subset of ImageNet validate the BOCL framework. We also reveal the performance accuracy of different sampling strategies when used to finetune a given CNN model. The code will be released.

Numerous methods have been proposed for person re-identification (Re-ID) with promising performances. While most of them neglect the matching efficiency which is crucial in real-world applications. Recently, several hashing based approaches have been developed, which consider the importance of matching speed in large-scale datasets. Despite the considerable efficiency of these traditional and deep learning based hashing methods, the concomitant matching accuracy reduction is unacceptable in practical application. Towards this end, we propose a novel deep hashing framework, namely Consistency-Preserving Deep Hashing (CPDH), aiming to bridge the gap between the effective high-dimensional feature and low-dimensional binary vector by focusing on the consistency preservation of hash code. First, CPDH designs a new hash structure to extract the hash code. Next, a noise consistency cost is proposed to improve robustness of both hash code and high-dimensional feature. Finally, a topology consistency cost is provided to maintain the ordinal relation between the high-dimensional feature space and Hamming space. Comprehensive experimental results on three widely-used benchmark datasets demonstrate the superior performance of proposed method as compared with existing state-of-the-art approaches.

Class-incremental learning (CIL) aims to incrementally learn a unified classifier for new classes emerging, which suffers from the catastrophic forgetting problem. To alleviate forgetting and improve the recognition performance, we propose a novel CIL framework, named the topological schemas model (TSM). TSM consists of a Gaussian mixture model arranged on 2D grids (2D-GMM) as the memory of the learned knowledge. To train the 2D-GMM model, we develop a novel competitive expectation-maximization (CEM) method, which contains a global topology embedding step and a local expectation-maximization fine-tuning step. Meanwhile, we choose the image samples of old classes that have the maximum posterior probability with respect to each Gaussian distribution as the episodic points. When finetuning for new classes, we propose the memory preservation loss (MPL) term to ensure episodic points still have maximum probabilities with respect to the corresponding Gaussian distribution. MPL preserves the distribution of 2D-GMM for old knowledge during incremental learning and alleviates catastrophic forgetting. Comprehensive experimental evaluations on two popular CIL benchmarks CIFAR100 and subImageNet demonstrate the superiority of our TSM.

With the continuous development of the national big data strategy and the construction of "Digital China", cities have been given new connotations and requirements. This paper takes the real estate price changes in Hainan Province as the background, collects relevant information and data, and makes relevant analysis and reasonable predictions for the housing price changes in Hainan Province. This paper comprehensively considers the prediction defects of MEA_BP neural network, Elman neural network and Wavelet neural network, and uses a combination algorithm based on entropy method to comprehensively consider the results of all prediction models. On the one hand, it can improve the prediction accuracy of the model, and on the other hand, it can reduce the sensitivity of the prediction results to changes in a certain factor, making the prediction results more reliable. This paper establishes a multiple and effective new era smart city prediction model, which provides a reference for building a new city database supported by a new generation of information technology.

The ability to incrementally learn new classes is crucial to the development of real-world artificial intelligence systems. In this paper, we focus on a challenging but practical few-shot class-incremental learning (FSCIL) problem. FSCIL requires CNN models to incrementally learn new classes from very few labelled samples, without forgetting the previously learned ones. To address this problem, we represent the knowledge using a neural gas (NG) network, which can learn and preserve the topology of the feature manifold formed by different classes. On this basis, we propose the TOpology-Preserving knowledge InCrementer (TOPIC) framework. TOPIC mitigates the forgetting of the old classes by stabilizing NG's topology and improves the representation learning for few-shot new classes by growing and adapting NG to new training samples. Comprehensive experimental results demonstrate that our proposed method significantly outperforms other state-of-the-art class-incremental learning methods on CIFAR100, miniImageNet, and CUB200 datasets., Accepted by CVPR 2020 (oral)

Semi-supervised learning is important and has become more widespread because obtaining labeled data is expensive and labor-intensive. In this paper, we focus on the challenging semi-supervised person Re-identification (ReID) task, which is a metric learning problem based on the assumption that unlabeled data is open-set. To address this problem, we propose the Transductive Semi-Supervised Metric Learning (TSSML) framework. In TSSML, we propose a graph-based transductive hard mining method for deeply mining hard triplets in unlabeled data and a degree-based relationship confidence scoring method for further reducing incorrect triplets. Moreover, we investigate the feature consistency loss (FCL) and adopt the curriculum learning strategy to improve the representation learning for semi-supervised ReID. Extensive experiments have been conducted on three large-scale ReID datasets and demonstrate the effectiveness of our TSSML framework.