Neuromorphic computing concept introduced by the Carver Mead in late 1980s. These chips are described as analog, digital mixed-mode VLSI (very-large-scale integration) system. This concept is analogous to human brain neural network, so we can call it an electronic brain. So far we know about the Artificial Neural Networks (ANN) which is based on mathematical calculations implemented as machine instructions. Unlike the von Neumann architecture, this system doesn’t need separate memory. Image shown below depicts the basic structure of the neuromorphic chip.
These chips have several advantages over conventional microchips.
- Low power consumption
- Can be used in real time pattern recognition applications
- Can reduce huge amount of data
Qualcomm and IBM have already started to manufacture neuromorphic chips. This technology will make the performance of pattern recognition application such as face-recognition, voice recognition, autonomous robots, natural language processing (NLP), weather predictions and etc.
Moreover, DigitalTonto has published that IBM has created new model according to macaque brain. This architecture has been designed by Dhamendra Modha who is a researcher at IBM. Modha’s network diagram of the macaque brain can be seen below. It contains main lobes of the human brain.
- Neuromorphic Chips Are Destined for Deep Learning—or Obscurity
- Neuromorphic Chips (MIT)
- IBM Has Created A Revolutionary New Model For Computing—The Human Brain
- Disrupted Humanity – Neuromorphic chips will help computers to think
- Neuromorphic Architectures (James Kempsell and Chris Radnovich)
- Neuromorphic Engineering / Computing by Kevin Espera