Towards Implementing Compression and Decompression Techniques on FPGA for Deep Neural Networks