INCREMENTAL LEARNING OF CHUNK DATA FOR ONLINE PATTERN CLASSIFICATION SYSTEMS

Incremental Learning of Chunk Data for   Online Pattern Classification Systems

Abstract:

      In this Project we approach a new technique called Chunk Incremental Principal Component Analysis which is the Enhancement of IPCA which is the enhancement Principal Component Analysis.

      This project presents a pattern classification system in which feature extraction and classifier learning are simultaneously carried out not only online but also in one pass where training samples are presented only once.

      For this purpose, we have extended incremental principal component analysis (IPCA) and some classifier models were effectively combined with it.

      However, there was a drawback in this approach that training samples must be learned one by one due to the limitation of IPCA. To overcome this problem, we propose another extension of IPCA called chunk IPCA in which a chunk of training samples is processed at a time.

      In the experiments, we evaluate the classification performance for several large-scale data sets to discuss the scalability of chunk IPCA under one-pass incremental learning environments.

      The experimental results suggest that chunk IPCA can reduce the training time effectively as compared with IPCA unless the number of input attributes is too large.

      We study the influence of the size of initial training data and the size of given chunk data on classification accuracy and learning time. We also show that chunk IPCA can obtain major eigenvectors with fairly good approximation.

Existing System:

·         Drawback of existing scheme was scalability in terms of the number of samples and the number of their attributes.

·         Training samples must be learned one by one due to the limitation of IPCA.

·         Existing system causes inefficiency in computations because the eigenvalue decomposition in IPCA must be applied to each training sample in the chunk.

Dis-advantages

·        The biggest problem is scalability. In our previous approach, a training sample must be learned one by one even if a chunk of training sample is available at a time.

·        This causes inefficiency in computations because the Eigen value decomposition in IPCA must be applied to each training sample in the chunk.

Proposed System:

·         We have proposed an extension of IPCA called chunk IPCA in which the update of an eigen space is completed by performing single eigenvalue decomposition.

·         We extended the incremental learning of a classifier such that it can also be learned with a chunk of training samples.

Advantages:

·         Proposed method possesses excellent scalability in learning time without sacrificing classification accuracy.

·         Proposed chunk IPCA gives a fairly good approximation to major eigenvectors with large eigenvalues.

Hardware Requirements:

§  PROCESSOR        :  PENTIUM IV 2.6 GHz

§  RAM                      :    512 MB DD RAM

§  MONITOR            :    15” COLOR

§  HARD DISK         :    20 GB

§  FLOPPY DRIVE   :   1.44 MB

§  CDDRIVE             :    LG 52X

§  KEYBOARD         :   STANDARD 102 KEYS

§  MOUSE                 :    3 BUTTONS

Software Requirements:

§  FRONT END                    :  Jdk1.5.0_1, Swing

§  OPERATING SYSTEM   :  Window’s XP