Final Year Project, B. Tech. C.S.E [2022]
-
Research based project for improving the efficiency of Artificial Neural Network.
-
Used few bio-inspired algorithms like PSO, Genetic Algorithm, etc. instead of back-propagation algorithm to find the best set of weights or the best model depending on the given dataset.
-
Distribution was done using PySpark on a cluster (DCS, CUSAT) to improve time taken on larger datasets.
-
Implemented parallel versions for the algorithms (multi-threading). Also compared accuracy vs time, time vs number of threads, etc. found an improvement from existing methods.
-
The code works with any classification data given as a csv file and builds the best model in a lesser amount of time while maintaining relatively good accuracy.