Jia-Lien Hsu

Discovering nontrivial repeating patterns in music data

A repeating pattern in music data is defined as a sequence of notes which appears more than once in a music object. The themes are a typical kind of repeating patterns. The themes and other nontrivial repeating patterns are important music features which can be used for both content-based retrieval of music data and music data analysis. In this paper, we propose two approaches for fast discovering nontrivial repeating patterns in music objects. In the first approach, we develop a data structure called correlative matrix and its associated algorithms for extracting the repeating patterns. In the second approach, we introduce a string-join operation and a data structure called RP-tree for the same purpose. Experiments are performed to compare these two approaches with others. The results are further analyzed to show the efficiency and the effectiveness of our approaches.

Efficient repeating pattern finding in music databases

In this paper, we propose an approach for the extraction of the repeating patterns in music objects. A repeating pattern is a sequence of notes which appears more than once in a music object. It is one of the most important music features which can be used for both content-based retrieval of music data and music data analysis. We propose a data structure called correlative matrix and its associated algorithms for extracting all repeating patterns in a music object. Experiments are also performed and the results are analyzed to show the efficiency and the effectiveness of our approach.

Efficient theme and non-trivial repeating pattern discovering in music databases

Proposes an approach for the fast discovery of all non-trivial repeating patterns in music objects. A repeating pattern is a sequence of notes which appears more than once in a music object. The longest repeating patterns in music objects are typically their themes. The themes and other non-trivial repeating patterns are important musical features which can be used both for content-based retrieval of music data and for music data analysis. We present a data structure called an RP-tree (repeating pattern tree) and its associated algorithms for the fast extraction of all non-trivial repeating patterns in a music object. Experiments are performed to compare this method with related approaches. The results are further analysed to show the efficiency and effectiveness of our approach.

An approximate string matching algorithm for content-based music data retrieval

An approach for content based music data retrieval is proposed. In this approach, thematic feature strings, such as melody strings, rhythm strings, and chord strings are extracted from the original music objects and treated as the meta data to represent their contents. The problem of content based music data retrieval is then transformed into the string matching problem. A new approximate string matching algorithm is also proposed which provides fault tolerance ability according to the music characteristics. To show the efficiency of the algorithm, a set of experiments are performed to compare with the agrep and the fgrep utility on both synthetic and real music data.

Query by music segments: an efficient approach for song retrieval

We present the techniques for retrieving songs by music segments. A music segment consists of a segment type and the associated beat and pitch information. The similarity measures for the beat and pitch are defined. Two index structures for music segments are proposed, in which the minimal and maximal values of the beat and pitch of the music segments are stored to aid the song retrieval process. Moreover, the threshold propagation functions are developed for efficient approximate searching. Experiments are performed to show the superiority of this approach.

The effectiveness study of various music information retrieval approaches

In this paper, we describe the Ultima project which aims to construct a platform for evaluating various approaches of music information retrieval. Two kinds of approaches are adopted in this project. These approaches differ in various aspects, such as representations of music objects, index structures, and approximate query processing strategies. For a fair comparison, we propose a measurement of the retrieval effectiveness by recall-precision curves with a scaling factor adjustment. Finally, the performance study of the retrieval effectiveness based on various factors of these approaches is presented.

Efficient Near Neighbor Searching Using Multi-Indexes for Content-Based Multimedia Data Retrieval

Many content-based multimedia data retrieval problems can be transformed into the near neighbor searching problem in multidimensional feature space. An efficient near neighbor searching algorithm is needed when developing a multimedia database system. In this paper, we propose an approach to efficiently solve the near neighbor searching problem. In this approach, along each dimension an index is constructed according to the values of feature points of multimedia objects. A user can pose a content-based query by specifying a multimedia query example and a similarity measure. The specified query example will be transformed into a query point in the multi-dimensional feature space. The possible result points in each dimension are then retrieved by searching the value of the query point in the corresponding dimension. The sets of the possible result points are merged one by one by removing the points which are not within the query radius. The resultant points and their distances from the query point form the answer of the query. To show the efficiency of our approach, a series of experiments are performed to compare with the related approaches.

Applying Under-Sampling Techniques and Cost-Sensitive Learning Methods on Risk Assessment of Breast Cancer

Breast cancer is one of the most common cause of cancer mortality. Early detection through mammography screening could significantly reduce mortality from breast cancer. However, most of screening methods may consume large amount of resources. We propose a computational model, which is solely based on personal health information, for breast cancer risk assessment. Our model can be served as a pre-screening program in the low-cost setting. In our study, the data set, consisting of 3976 records, is collected from Taipei City Hospital starting from 2008.1.1 to 2008.12.31. Based on the dataset, we first apply the sampling techniques and dimension reduction method to preprocess the testing data. Then, we construct various kinds of classifiers (including basic classifiers, ensemble methods, and cost-sensitive methods) to predict the risk. The cost-sensitive method with random forest classifier is able to achieve recall (or sensitivity) as 100 %. At the recall of 100 %, the precision (positive predictive value, PPV), and specificity of cost-sensitive method with random forest classifier was 2.9 % and 14.87 %, respectively. In our study, we build a breast cancer risk assessment model by using the data mining techniques. Our model has the potential to be served as an assisting tool in the breast cancer screening.

A cross-modal method of labeling music tags

In this paper, we discuss various features of music objects in two kinds of domain. Among these features, Mel-frequency cepstral coefficients (MFCCs) are further discussed and described by Gaussian mixture model (GMM). Also, the similarity between GMMs are investigated accordingly. Then, we employ the multimedia graph as a cross-modal method to associate MFCCs and genre tags of music objects. By applying link analysis algorithm in the graph, we label appropriate genre tags for target music objects. Also, we perform experiments to show performance, effectiveness, and parameter setting of our approach.

Constraint-based playlist generation by applying genetic algorithm

In this paper, we propose a formal model of playlist generation and define three types of constraints, i.e., parameter-specified constraints, derived constraints, and user-defined constraints. Some constraints can be derived through user behaviour and user interactions. Given a set of constraints, we apply a genetic algorithm to generate a playlist which optimizes the number of matched constraints. We also implement our prototype, and perform experiments to show the feasibility and effectiveness of prototype.