While a large body of research has formally identified apolipoprotein E (APOE) as a major genetic risk marker for Alzheimer’s disease, accumulating evidence supports the notion that other risk markers may exist. The ...
详细信息
Vehicle re-identification (reID) aims at identifying vehicles across different non-overlapping cameras views. The existing methods heavily relied on well-labeled datasets for ideal performance, which inevitably causes...
详细信息
Deep subspace clustering (DSC) with the auto-encoder and self-expression layer is of great concern due to encouraging performance. However, existing methods usually adopt a “single-task” strategy based on a single d...
详细信息
ISBN:
(纸本)9781665423991
Deep subspace clustering (DSC) with the auto-encoder and self-expression layer is of great concern due to encouraging performance. However, existing methods usually adopt a “single-task” strategy based on a single dataset, without considering other related tasks or data. As such, they cannot discover other useful information to improve the clustering task. Besides, the local structure preservation of the latent codes in mapping is usually ignored. In this paper, we therefore present an effective “multi-task” strategy via the self-supervised data augmentation, and propose a new end-to-end trainable Triplet Deep Subspace Clustering Network (TDSC-net). Specifically, TDSC-net firstly generates triplet data (i.e., anchor, positive and negative data) from input data by a spectral clustering module and a self-supervised data augmentation module. This can enable it to inherit the merits of self-supervised learning and multitask learning implicitly. After that, TDSC-net builds a triplet deep autoencoder network with a self-expression layer, which takes the triplet data as input, where they share the common network layers (i.e., autoencoder and self-expression layers) over the triple tasks for complementary learning and mutual supervision. A triplet loss is also included to retain the local information of deep latent codes, which will also benefit the self-expression. Furthermore, TDSC-net separates the self-expression layer from decoding process to improve the efficiency of reconstruction. Extensive results on several public datasets demonstrate the effectiveness of our triplet-task DSC strategy.
Dictionary learning (DL) is powerful for representation learning, while it fails to capture the deep hierarchical information hidden in data. In this paper, we propose a new generalized end-to-end mulita-layer represe...
详细信息
ISBN:
(数字)9781728183169
ISBN:
(纸本)9781728183176
Dictionary learning (DL) is powerful for representation learning, while it fails to capture the deep hierarchical information hidden in data. In this paper, we propose a new generalized end-to-end mulita-layer representation learning architecture referred to as Multi-layer Dictionary Pair Learning Network (MDPL-net) for the deep sparse and hierarchical representation of images. To enable MDPL-net to conduct accurate classification, MDPL-net clearly integrates the skip connection end-to-end network and multi-layer deep sparse dictionary learning into a unified architecture. The representation learning module has several hidden DL blocks, where each hidden DL block has a dictionary pair learning (DPL) layer, a batch-norm layer and an activation function layer, and the DL blocks are connected in a feed-forward manner. To further improve the information flow and maintain the privileged features between different DL blocks, a novel skip dense connectivity pattern is deployed between hidden DL blocks, which can obtain more stable and discriminative features. The DPL layer jointly formulates the discriminative synthesis dictionary and analysis dictionary by minimizing reconstruction error within each batch over the feature maps from front layers. Extensive results on benchmark databases demonstrate the effectiveness of MDPL-net for discriminative representation and robust image classification.
Compared with univariate time series clustering, multivariate time series (MTS) clustering has become a challenging research topic on the data mining of time series. In this paper, a novel model-based approach Adaptiv...
详细信息
ISBN:
(数字)9781728159287
ISBN:
(纸本)9781728159294
Compared with univariate time series clustering, multivariate time series (MTS) clustering has become a challenging research topic on the data mining of time series. In this paper, a novel model-based approach Adaptive State Continuity-Based Sparse Inverse Covariance Clustering (ASCSICC) is proposed for MTS clustering. Specifically, the state continuity is introduced to make the traditional Gaussian mixture model (GMM) applicable to time series clustering. To prevent overfitting, the alternating direction method of multipliers (ADMM) is applied to optimize the parameter of GMM inverse covariance. In addition, the proposed approach simultaneously segments and clusters the time series. Technically, first, the adaptive state continuity is estimated based on the distance similarity of adjacent time series data. Then, a dynamic programming algorithm of cluster assignment by adaptive state continuity is taken as the E-step, and the ADMM for solving sparse inverse covariance is taken as the M-step. E-step and M-step are combined into an Expectation-Maximization (EM) algorithm to conduct the clustering process. Finally, we show the effectiveness of the proposed approach by comparing the ASC-SICC with several state-of-the-art approaches in experiments on two datasets from real applications.
Nowadays, vision-based computing tasks play an important role in various real-world applications. However, many vision computing tasks, e.g. semantic segmentation, are usually computationally expensive, posing a chall...
详细信息
This paper attacks the challenging problem of video retrieval by text. In such a retrieval paradigm, an end user searches for unlabeled videos by ad-hoc queries described exclusively in the form of a natural-language ...
详细信息
In order to better understand the stochastic dynamic features of signalized traffic networks, we propose a Markov traffic model to simulate the dynamics of traffic link flow density for signalized urban traffic networ...
详细信息
In order to better understand the stochastic dynamic features of signalized traffic networks, we propose a Markov traffic model to simulate the dynamics of traffic link flow density for signalized urban traffic networks with demand uncertainty. In this model, we have four different state modes for the link according to different congestion levels of the link. Each link can only be in one of the four link state modes at any time, and the transition probability from one state to the other state is estimated by Bayesian estimation based on the distributions of the dynamic traffic flow densities, and the posterior probabilities. Therefore, we use a first-order Markov Chain Model to describe the dynamics of the traffic flow evolution process. We illustrate our approach for a small traffic network. Compared with the data from the microscopic traffic simulator SUMO, the proposed model can estimate the link traffic densities accurately and can give a reliable estimation of the uncertainties in the dynamic process of signalized traffic networks.
Recent works have managed to learn cross-lingual word embeddings (CLWEs) in an unsupervised manner. As a prominent unsupervised model, generative adversarial networks (GANs) have been heavily studied for unsupervised ...
详细信息
Recent works have managed to learn cross-lingual word embeddings (CLWEs) in an unsupervised manner. As a prominent unsupervised model, generative adversarial networks (GANs) have been heavily studied for unsupervised CLWEs learning by aligning the embedding spaces of different languages. Due to disturbing the embedding distribution, the embeddings of low-frequency words (LFEs) are usually treated as noises in the alignment process. To alleviate the impact of LFEs, existing GANs based models utilized a heuristic rule to aggressively sample the embeddings of high-frequency words (HFEs). However, such sampling rule lacks of theoretical support. In this paper, we propose a novel GANs based model to learn cross-lingual word embeddings without any parallel resource. To address the noise problem caused by the LFEs, some perturbations are injected into the LFEs for offsetting the distribution disturbance. In addition, a modified framework based on Cramér GAN is designed to train the perturbed LFEs and the HFEs jointly. Empirical evaluation on bilingual lexicon induction demonstrates that the proposed model outperforms the state-of-the-art GANs based model in several language pairs.
Due to the lack of theoretical knowledge and practical experience, the university students have many deficiencies in contribution. We conducted semi-structured interviews with 28 undergraduate and postgraduate student...
Due to the lack of theoretical knowledge and practical experience, the university students have many deficiencies in contribution. We conducted semi-structured interviews with 28 undergraduate and postgraduate students of Nanjing Agricultural University. At the same time, we used descriptive statistics and cross-analysis methods to analyse the motivations of college students' initial contribution and the influencing factor of their choice. Based on this, this paper puts forward some suggestions for the initial submission of college students, and the journals are also provided for reference in the selection of manuscripts.
暂无评论