Jia Rong

Acoustic feature selection for automatic emotion recognition from speech

Emotional expression and understanding are normal instincts of human beings, but automatical emotion recognition from speech without referring any language or linguistic information remains an unclosed problem. The limited size of existing emotional data samples, and the relative higher dimensionality have outstripped many dimensionality reduction and feature selection algorithms. This paper focuses on the data preprocessing techniques which aim to extract the most effective acoustic features to improve the performance of the emotion recognition. A novel algorithm is presented in this paper, which can be applied on a small sized data set with a high number of features. The presented algorithm integrates the advantages from a decision tree method and the random forest ensemble. Experiment results on a series of Chinese emotional speech data sets indicate that the presented algorithm can achieve improved results on emotional recognition, and outperform the commonly used Principle Component Analysis (PCA)/Multi-Dimensional Scaling (MDS) methods, and the more recently developed ISOMap dimensionality reduction method.

Mining permission patterns for contrasting clean and malicious android applications

Abstract An Android application uses a permission system to regulate the access to system resources and users’ privacy-relevant information. Existing works have demonstrated several techniques to study the required permissions declared by the developers, but little attention has been paid towards used permissions. Besides, no specific permission combination is identified to be effective for malware detection. To fill these gaps, we have proposed a novel pattern mining algorithm to identify a set of contrast permission patterns that aim to detect the difference between clean and malicious applications. A benchmark malware dataset and a dataset of 1227 clean applications has been collected by us to evaluate the performance of the proposed algorithm. Valuable findings are obtained by analyzing the returned contrast permission patterns.

An effective privacy preserving algorithm for neighborhood-based collaborative filtering

As a popular technique in recommender systems, Collaborative Filtering (CF) has been the focus of significant attention in recent years, however, its privacy-related issues, especially for the neighborhood-based CF methods, cannot be overlooked. The aim of this study is to address these privacy issues in the context of neighborhood-based CF methods by proposing a Private Neighbor Collaborative Filtering (PNCF) algorithm. This algorithm includes two privacy preserving operations: Private Neighbor Selection and Perturbation. Using the item-based method as an example, Private Neighbor Selection is constructed on the basis of the notion of differential privacy, meaning that neighbors are privately selected for the target item according to its similarities with others. Recommendation-Aware Sensitivity and a re-designed differential privacy mechanism are introduced in this operation to enhance the performance of recommendations. A Perturbation operation then hides the true ratings of selected neighbors by adding Laplace noise. The PNCF algorithm reduces the magnitude of the noise introduced from the traditional differential privacy mechanism. Moreover, a theoretical analysis is provided to show that the proposed algorithm can resist a KNN attack while retaining the accuracy of recommendations. The results from experiments on two real datasets show that the proposed PNCF algorithm can obtain a rigid privacy guarantee without high accuracy loss.

Acoustic Features Extraction for Emotion Recognition

In the last decade, the efforts of spoken language processing have achieved significant advances, however, the work with emotional recognition has not progressed so far, and can only achieve 50% to 60% in accuracy. This is because a majority of researchers in this field have focused on the synthesis of emotional speech rather than focusing on automating human emotion recognition. Many research groups have focused on how to improve the performance of the classifier they used for emotion recognition, and few work has been done on data pre-processing, such as the extraction and selection of a set of specifying acoustic features instead of using all the possible ones they had in hand. To work with well-selected acoustic features does not mean to delay the whole job, but this will save much time and resources by removing the irrelative information and reducing the high-dimension data calculation. In this paper, we developed an automatic feature selector based on a RF2TREE algorithm and the traditional C4.5 algorithm. RF2TREE applied here helped us to solve the problems that did not have enough data examples. The ensemble learning technique was applied to enlarge the original data set by building a bagged random forest to generate many virtual examples, and then the new data set was used to train a single decision tree, which selects the most efficient features to represent the speech signals for the emotion recognition. Finally, the output of the selector was a set of specifying acoustic features, produced by RF2TREE and a single decision tree.

Personality Differences and Hotel Web Design Study Using Targeted Positive and Negative Association Rule Mining

As people have unique tastes, the way to satisfy a small group of targeted customers or to be generic to meet most peoples preference has been a traditional question to many fashion designers and website developers. This study examined the relationship between individuals’ personality differences and their web design preferences. Each individuals personality is represented by a combination of five traits, and 15 website design-related features are considered to test the users’ preference. We introduced a data mining technique called targeted positive and negative association rule mining to analyze a dataset containing the survey results collected from undergraduate students. The results of this study not only suggest the importance of providing specific designs to attract individual customers, but also provide valuable input on the Big Five personality traits in their entirety.

Contrasting Permission Patterns between Clean and Malicious Android Applications

The Android platform uses a permission system model to allow users and developers to regulate access to private information and system resources required by applications. Permissions have been proved to be useful for inferring behaviors and characteristics of an application. In this paper, a novel method to extract contrasting permission patterns for clean and malicious applications is proposed. Contrary to existing work, both required and used permissions were considered when discovering the patterns. We evaluated our methodology on a clean and a malware dataset, each comprising of 1227 applications. Our empirical results suggest that our permission patterns can capture key differences between clean and malicious applications, which can assist in characterizing these two types of applications.

An Analysis on Human Personality and Hotel Web Design: a Kohonen Network Approach

Having an eye catching and attractive website could help hotels to compete in the vigorous online market. This study attempts to examine the relationship between human personality and the web design preferences. Kohonen Networks were adopted to cluster people with similar personality characteristics and identify their differences on web design preferences. Empirical results indicated people with similar personality traits have similar design preferences. For example, to attract those who got high scores in agreeableness, conscientiousness and openness but low score in neuroticism, a web page should start with a language selection page with introductory movie, one large image on the web page showing hotel interior design with hotel guest in the photo, and with background music.

Sharing sensitive medical data sets for research purposes - A case study

For medical research purposes, having access to large sets of data, often from various regions, improves statistical outcomes of analysis. However, patient data is usually considered to be sensitive and access to it is restricted by law and regulation. This paper employs privatization techniques which enable sharing of sensitive data. We demonstrate a case study on four medical data sets.