BIO-INFORMATICS DATA CLASSIFICATION USING KNN ON FPGA BOARD

Abstract

Bioinformatics data is treated as high-dimensional data by nature, which requires great computational demands. Scientists around the world have proposed many computing solutions such as Field-Programmable Gate Arrays (FPGAs) and Graphics Processing Units (GPUs). In order to meet these high computational demands, a feature wherein we can alter some specific areas in the chip proves to be very helpful and resourceful, from which FPGAs are benefiting. FPGAs enable flexible, reconfigurable computing, as most enable the user to reprogram the hardware circuit with different logic functions. Applying classification machine learning algorithms to the bioinformatics data set on a conventional PC to get the desired result proves to be a very time-consuming task. This is where the FPGAs come into the picture and help in drastically reducing this computational time. In this project, we have implemented the above approach using an FPGA board and executed its software-based implementation on a CPU to compare them on the grounds of timing. The hardware implementation of the algorithm is done using Verilog and for the software-based implementation we have used Python. Furthermore, in this project we have done a comparative analysis by adopting different sorting technique which plays a vital role in the KNN classification algorithm implemented. These algorithms also are one of the pivotal factors for the speed and hardware utilization requirement for implementing the algorithm on the FPGA board.