دانلود کتاب اصول و سازماندهی محاسبات نورومورفیک

As the end of Moore’s law seems closer than ever, computer scientists have been exploring to build machines as complex and efficient as our brain, dealing with power density and clock frequency challenges of the conventional architecture. Our brain works entirely differently compared to traditional von Neumann architecture. There are many secrets behind how the human brain works. We know that it distributes computation and memory among more than 100 billion biological neurons, and each of them is connected with thousands of others via synapses. Neurons communicate with each other through spikes (i.e., short electrical pulses). The brain is a powerful computation system that helps us survive, adapt, and predict, while consuming tens of watts. Brain-inspired or neuromorphic computing is a biologically inspired approach created from highly connected neurons to model neuroscience theories and solve machine learning problems. The term neuromorphic was first introduced by Carver Mead in 1990, where it referred to very-large-scale integration (VLSI) with analog components to mimic biological neural systems. Such systems can be categorized into non-spiking and spiking approaches. First, the non-spiking approach is referred to as the implementation of traditional artificial neural networks (ANNs) which aims to improve the throughput over the power consumption (or acceleration purpose). In recent years, ANNs have shown a remarkable improvement in terms of accuracy for large-scale visual/auditory recognition and classification tasks. Notably, the convolution neural network (CNN) and recurrent neural network (RNN) have shown to be promising tools for a wide range of applications such as image, video, and speech. They are typically trained by using graphic processing units (GPUs) or on the cloud side. The state-of-the-art neural networks tend to increase their number of layers and size (i.e., deep learning). However, this leads to challenges for hardware systems in terms of computation, memory, and communication resources.