Zhihua Wei

Mixed Pooling for Convolutional Neural Networks

Convolutional Neural Network (CNN) is a biologically inspired trainable architecture that can learn invariant features for a number of applications. In general, CNNs consist of alternating convolutional layers, non-linearity layers and feature pooling layers. In this work, a novel feature pooling method, named as mixed pooling, is proposed to regularize CNNs, which replaces the deterministic pooling operations with a stochastic procedure by randomly using the conventional max pooling and average pooling methods. The advantage of the proposed mixed pooling method lies in its wonderful ability to address the over-fitting problem encountered by CNN generation. Experimental results on three benchmark image classification datasets demonstrate that the proposed mixed pooling method is superior to max pooling, average pooling and some other state-of-the-art works known in the literature.

N-grams based feature selection and text representation for Chinese Text Classification

In this paper, text representation and feature selection strategies for Chinese text classification based on n-grams are discussed. Two steps feature selection strategy is proposed which combines the preprocess within classes with the feature selection among classes. Four different feature selection methods and three text representation weights are compared by exhaustive experiments. Both C-SVC classifier and Naive bayes classifier are adopted to assess the results. All experiments are performed on Chinese corpus TanCorpV1.0 which includes more than 14,000 texts divided in 12 classes. Our experiments concern: (1) the performance comparison among different feature selection strategies: absolute text frequency, relative text frequency, absolute n-gram frequency and relative n-gram frequency; (2) the comparison of the sparseness and feature correlation in the “text by feature” matrices produced by four feature selection methods; (3) the performance comparison among three term weights: 0/1 logical value, n-gr...

A Framework of Video Coding for Compressing Near-Duplicate Videos

With the development of multimedia technique and social network, the amount of videos has grown rapidly, which brings about an increasingly substantial percentage of Near-Duplicate Videos (NDVs). It has been a hot research topic to retrieve NDVs for a number of applications such as copyright detection, Internet video ranking, etc. However, there exist a lot of redundancies in NDVs, and to the best of our knowledge it is an untouched research area on how to efficiently compress NDVs in a joint manner. In this work, a novel video coding framework is proposed to effectively compress NDVs by making full use of the relevance among them. Experimental results demonstrate that a significant storage saving can be achieved by the proposed NDV coding framework.

Semi-supervised multi-label image classification based on nearest neighbor editing

Semi-supervised multi-label classification has been applied to many real-world applications such as image classification, document classification and so on. In semi-supervised learning, unlabeled samples are added to the training set for enhancing the classification performance, however, noises are introduced simultaneously. In order to reduce this negative effect, the nearest neighbor data editing technique is introduced to semi-supervised multi-label classification, and thus an algorithm named Multi-Label Self-Training with Editing (MLSTE) is proposed in this work. The proposed algorithm is able to solve the uncertainty problem in semi-supervised multi-label classification to some extent, by improving the performance of determining the label number and selecting confident samples during the course of semi-supervised learning. Extensive experimental results on several benchmark datasets have been carried out to verify the effectiveness of the proposed MLSTE algorithm.

From Principal Curves to Granular Principal Curves

Principal curves arising as an essential construct in dimensionality reduction and data analysis have recently attracted much attention from theoretical as well as practical perspective. In many real-world situations, however, the efficiency of existing principal curves algorithms is often arguable, in particular when dealing with massive data owing to the associated high computational complexity. A certain drawback of these constructs stems from the fact that in several applications principal curves cannot fully capture some essential problem-oriented facets of the data dealing with width, aspect ratio, width change, etc. Information granulation is a powerful tool supporting processing and interpreting massive data. In this paper, invoking the underlying ideas of information granulation, we propose a granular principal curves approach, regarded as an extension of principal curves algorithms, to improve efficiency and achieve a sound accuracy-efficiency tradeoff. First, large amounts of numerical data are granulated into

Bayesian rule based fast TU depth decision algorithm for high efficiency video coding

C

Optimal stopping theory based fast coding tree unit decision for high efficiency video coding

intervals-information granules developed with the use of fuzzy C-means clustering and the two criteria of information granulation, which significantly reduce the amount of data to be processed at the later phase of the overall design. Granular principal curves are then constructed by determining the upper and the lower bounds of the interval data. Finally, we develop an objective function using the criteria of information confidence and specificity to evaluate the granular output formed by the principal curves. We also optimize the granular principal curves by adjusting the level of information granularity (the number of clusters), which is realized with the aid of the particle swarm optimization. A number of numeric studies completed for synthetic and real-world datasets provide a useful quantifiable insight into the effectiveness of the proposed algorithm.

Research on highly consumable platform for business analytics

The latest video coding standard high efficiency video coding (HEVC) has made a significant progress in compression efficiency than previous standard H.264/advanced video coding (AVC) while it has led to a tremendous increase in encoding computations. Recently, a Bayesian model based transform unit (TU) depth decision approach has been designed to accelerate TU depth decision, which requires numerous variance computations. In this work, a novel relevant feature based Bayesian model is proposed for fast TU depth decision. Experimental results demonstrate that the best performance is achieved while the depths of upper TU, left TU and co-located TU are all taken into considerations. Moreover, as compared with previous research, the proposed algorithm reduces much more encoding computations while keeping the video quality and compression efficiency more or less intact.

Document-Level Sentiment Classification Based on Behavior-Knowledge Space Method

High Efficiency Video Coding (HEVC) is the most recent video coding standard aiming to further reduce the bitrate over 50% as compared to the state-of-the-art H.264/Advanced Video Coding under the same visual quality. In order to achieve this, a number of advanced coding techniques have been adopted in HEVC, including the quadtree structure of Coding Unit (CU), Prediction Unit (PU) and Transform Unit (TU), etc. However, these coding techniques lead to a tremendous increase in HEVC encoding computations. In order to reduce the HEVC encoding computational complexity, the optimal stopping theory is employed herein to design an efficient algorithm to optimize the decision making process when choosing the best coding parameters of CU, PU and TU. Extensive comparative experimental results are performed by the proposed algorithm and another two recent works, which demonstrate that the proposed algorithm is very efficient and better in reducing the HEVC encoding computations while keeping the video quality and compression efficiency almost intact.

Online Object Tracking via Collaborative Model within the Cascaded Feedback Framework

Business analytics is an important means of gaining new insights and understanding of business performance. Data mining technology provides information needed by business analytics based on data. But due to the barriers of skill, resource and cost, business analytics is hard to carry out and put into practice for small and even medium sized enterprises. This paper presents a novel concept named Highly Consumable Business Analytics (HCBA) aiming to improve consumability of business analytics. We discuss the ways to simplify the use of data mining algorithms, establish end-to-end integrated lifecycle management of model and ultimately wrap models into web service to provide model-wise services. We propose a general reference architecture of HCBA. And a prototype is designed and implemented based on this architecture.