MODIFIED AIS-BASED CLASSIFIER FOR MUSIC GENRE CLASSIFICATION
Automating human capabilities for classifying different genre of songs is a difficult task. This has led to various studies that focused on finding solutions to solve this problem. Analyzing music contents (often referred as content- based analysis) is one of many ways to identify and group sim...
| Main Author: | |
|---|---|
| Format: | Conference or Workshop Item |
| Published: |
2010
|
| Subjects: | |
| Online Access: | http://eprints.utem.edu.my/156/ http://eprints.utem.edu.my/156/1/Modified_AIS-Based_Classifier_for_Music_Genre_Classification.pdf |
| Summary: | Automating human capabilities for classifying different
genre of songs is a difficult task. This has led to various
studies that focused on finding solutions to solve this
problem. Analyzing music contents (often referred as content-
based analysis) is one of many ways to identify and
group similar songs together. Various music contents, for
example beat, pitch, timbral and many others were used
and analyzed to represent the music. To be able to manipulate
these content representations for recognition: feature
extraction and classification are two major focuses of
investigation in this area. Though various classification
techniques proposed so far, we are introducing yet another
one. The objective of this paper is to introduce a possible
new technique in the Artificial Immune System (AIS)
domain called a modified immune classifier (MIC) for
music genre classification. MIC is the newest version of
Negative Selection Algorithm (NSA) where it stresses the
self and non-self cells recognition and a complementary
process for generating detectors. The discussion will detail
out the MIC procedures applied and the modified part
in solving the classification problem. At the end, the results
of proposed framework will be presented, discussed
and directions for future work are given. |
|---|