XNOR-Net-Binary-Connect

Two hardware implementation of a feed-forward Convolutional Neural Network

XNOR-Net : faster execution due to the replacement of vector-matrix multiplication to “XNOR + Popcount” operation
Binary Connect : faster execution due to the presence of accumulator
For the current scope of the project, we consider MNIST dataset which consists of 10 classes of images i.e. 0 to 9.
Once the training of the neural network is done using any machine, our hardware design focuses on achieving speedup in the classification phase
The weights learned by the neural network will be inserted in the module using a BRAM to the feed forward network
These weights along with the images to be classified would be used as an input to the feed forward network on an FPGA
The scope of our project is only defined for the implementation of the fully connected layer after the convolutions have been done

For more information click here

Name		Name	Last commit message	Last commit date
Latest commit History 6 Commits
python_sims		python_sims
Common.vhd		Common.vhd
Full_Adder.vhd		Full_Adder.vhd
Latch.vhd		Latch.vhd
Latch_FF.vhd		Latch_FF.vhd
Popcount.vhd		Popcount.vhd
Popcount_32.vhd		Popcount_32.vhd
Popcount_4.vhd		Popcount_4.vhd
Popcount_8.vhd		Popcount_8.vhd
README.md		README.md
Ripple_Carry_Adder.vhd		Ripple_Carry_Adder.vhd
XNOR-NET.pdf		XNOR-NET.pdf
XNOR-NET.pptx		XNOR-NET.pptx
accumulator.vhd		accumulator.vhd
accumulator_ard.vhd		accumulator_ard.vhd
accumulator_tb.vhd		accumulator_tb.vhd
accumulator_tb_ard.vhd		accumulator_tb_ard.vhd
scalar_multiplier_ard.vhd		scalar_multiplier_ard.vhd
scalar_multiplier_last.vhd		scalar_multiplier_last.vhd
scalar_multiplier_sfixed.vhd		scalar_multiplier_sfixed.vhd
scalar_multiplier_v.v		scalar_multiplier_v.v
scalar_multiplier_v_32b.v		scalar_multiplier_v_32b.v
scalar_multiplier_v_area.v		scalar_multiplier_v_area.v
scalar_multiplier_v_tb.v		scalar_multiplier_v_tb.v
tb_Full_Adder.vhd		tb_Full_Adder.vhd
tb_SR_Latch.vhd		tb_SR_Latch.vhd
tb_XNOR_net.vhd		tb_XNOR_net.vhd
tb_binary_connect.vhd		tb_binary_connect.vhd
tb_popcount.vhd		tb_popcount.vhd
tb_popcount_32.vhd		tb_popcount_32.vhd
tb_xnor_popcount_32.vhd		tb_xnor_popcount_32.vhd
tb_xnor_popcount_8.vhd		tb_xnor_popcount_8.vhd
xnor_popcount_32.vhd		xnor_popcount_32.vhd
xnor_popcount_8.vhd		xnor_popcount_8.vhd

prateek22sri/XNOR-net-Binary-connect