We introduce BinaryNet, a method which trains DNNs with binary weights and activations when computing parameters’ gradient. We show that it is possible to train a Multi Layer Perceptron (MLP) on MNIST and ConvNets on CIFAR-10 and SVHN with BinaryNet and achieve nearly state-of-the-art results. At run-time, BinaryNet drastically reduces memory usage and replaces most multiplications by 1-bit exc...

Transistor sizing is very important for determination of the circuit performance. As a result for providing fair evaluation, an optimal size of transistor is necessary. Genetic algorithm that is capable of reduction of search problem complexity uses the transistor sizing which is originally a kind of search problem in the large multidimensional search space for energy consumption optimization. ...

In recent year, power dissipation is one of the biggest challenges in VLSI design. Multipliers are the main sources of power dissipation in DSP blocks. In this project various types of full adders design are performed. Different techniques are used for low power in full adders. The design and power comparison of the low power multiplier using different types of full adder adders units are analy...

All-optical devices have demonstrated a broad range of applications in the communication field. These serve as fundamental building blocks sophisticated integrated circuits. By integrating these into fields such signal processing, chip design, and network computations, it is possible to achieve much more efficient device. This paper describes design all-optical logic gates NAND, NOR, XNOR using...

The COVID-19 pandemic is spreading around the world causing more than 177 million cases and over 3.8 deaths according to European Centre for Disease Prevention Control. virus has devastating effects on economies, health, well-being of worldwide population. Due high increase in daily cases, available number test kits under-developed countries scarce. Hence, it vital implement an effective screen...

In-memory computing (IMC) has been widely accepted to be an effective method improve energy efficiency. To realize IMC, operands in static random-access memory (SRAM) are stored columns, which contradicts SRAM write patterns and requires additional data movement. In this paper, 8T array with configurable word lines is proposed, where the arranged rows, following traditional storage pattern, the...

Extracting an effective facial feature representation is the critical task for automatic expression recognition system. Local Binary Pattern (LBP) known to be a popular texture recognition. However, only few approaches utilize relationship between local neighborhood pixels itself. This paper presents Hybrid Texture Descriptor (HLTD) which derived from logical fusion of Neighborhood XNOR Pattern...

In this work we introduce a binarized deep neural network (BDNN) model. BDNNs are trained using a novel binarized back propagation algorithm (BBP), which uses binary weights and binary neurons during the forward and backward propagation, while retaining precision of the stored weights in which gradients are accumulated. At test phase, BDNNs are fully binarized and can be implemented in hardware...

The Energy dissipation in conventional CMOS circuits can be minimized through adiabatic technique. By adiabatic technique dissipation in PMOS network can be minimized and some of energy stored at load capacitance can be recycled instead of dissipated as heat. But the adiabatic technique is highly dependent on parameter variation. The intension of the paper is to implement combinational and sequ...