Design & implementation of machine learning algorithm in (2)
- 1. Submitted To: Dr. Neelima Gupta Dr. Sandhya Aneja Dr. Vipin Kumar Submitted By: Saurabh Kumar Chaudhary M.Sc. (Computer Science) Roll No. 20829
- 2. Message passing interface(MPI) • MPI is an industrial standard that specifies library routines needed for writing message passing programs. • MPI allows the development of scalable portable message passing programs • MPI uses a library approach to support parallel programming • MPI programs are compiled with regular compiler (e.g. gcc) and linked with an MPI library • No shared memory
- 3. BASIC MPI ROUTINES • MPI_INIT : INITIATE AN MPI COMPUTATION. • MPI_FINALIZE : TERMINATE A COMPUTATION. • MPI_COMM_SIZE : DETERMINE NUMBER OF PROCESSES. • MPI_COMM_RANK : DETERMINE MY PROCESS IDENTIFIER. • MPI_SEND : SEND A MESSAGE • MPI_RECV : RECEIVE A MESSAGE.
- 4. PROBLEMS WITH SERIAL PROGRAMMING • MOORE'S LAW AND ITS LIMITS • CHIP PERFORMANCE DOUBLES EVERY 18-24 MONTHS • LIMITS OF SERIAL COMPUTING • HEATING ISSUES • LIMIT TO TRANSMISSIONS SPEEDS.
- 5. WHY MPI ? • Thus need development of explicit parallel algorithms that are based on a fundamental understanding of the parallelism inherent in a problem, and exploiting that parallelism with minimum interaction/communication between the parallel parts • MPI provides a powerful, efficient, and portable way to express parallel programs • MPI was explicitly designed to enable libraries… • Which may eliminate the need for many users to learn (much of) MPI
- 6. MACHINE LEARNING • Field of study that gives computers the ability to learn without being explicitly programmed. • Machine learning explores the study and construction of algorithms/model that can learn from and make predictions on data. • A computer program is said to learn from experience E with respect to some class of tasks t and Performance measure P if its performance at tasks in T, as measured by P, improves with experience E .
- 7. CLASSIFICATION OF MACHINE LEARNING •Supervised learning •Classification •Regression •Unsupervised leaning •Clustering
- 8. K-NN : K NEAREST NEIGHBOR SEARCH ALGORITHM • Searches the feature space for the k training instances that are closest to the unknown instance (test tuple) • Closeness/similarity : in terms of distance metric.
- 9. PROBLEMS WITH SERIAL K-NN: • Time complexity • Sensitive to the local structure of the data • Curse of dimensionality.
- 10. SOLUTION: • Use parallel approach to search K-NN of the query instance.
- 11. WORK DONE • CNN (CONDENSED NEAREST NEIGHBOR) RULE. • SNN(SELECTIVE NEAREST NEIGHBOR) RULE. • RNN(REDUCED NEAREST NEIGHBOR )RULE. • Repeated ENN(EDITED NEAREST NEIGHBOR )RULE. • IB2 & IB3 (INSTANCE BASED) ALGORITHM.
- 12. OUR PROPOSED APPROACH •Pre-processing step • Perform clustering process on training set to divide it into p mutually exclusive partition {P1,P2,…,Pp}, where p is number of process . • Create the Representative Instance to represent each partition
- 13. STEP-II For i=1 to p • Apply k-means approach • Evaluate nearest neighbor similarity with representative instance(centroid) of each partition. • Perform • Competence Enhancement – Repeated Wilson Editing Rule (noise removal) • Competence Preservation (removal of superfluous instance) Update the centroid of the cluster. • Repeat step I & II until number of instances in the selected one partition >=k.
- 14. STEP-III •Take a test instance . •Select the partition whose R.I is closest to test instance. •Apply the majority rule. •Select the class label who has majority for the test instance.
- 15. UPDATION OF TRAINING SET • When the similarity value of the new test instance with the R.I. of the different partition exceeds the max radius value which we store during the pre-processing step. • Update the R.I. of that partition only, which is closed to the new test instance.
- 17. TIME COMPLEXITY (Single Time Investment) In each iteration • I-STEP (PRE PROCESSING STEP) : O(nd/p) • II-STEP : O(ndk/p) • IN THE LAST ITERATION III-STEP : o(kd)
- 18. FUTURE WORK •Selection of initial centroid in clustering in order to get speed up. •Selection of k •Number of processor
- 19. THANK YOU…