Abstract
Deep learning (DL) has been considered as a breakthrough technique in the field of artificial intelligence and machine learning. Conceptually, it relies on a many-layer network that exhibits a hierarchically non-linear processing capability. Some DL architectures such as deep neural networks, deep belief networks and recurrent neural networks have been developed and applied to many fields with incredible results, even comparable to human intelligence. However, many researchers are still sceptical about its true capability: can the intelligence demonstrated by deep learning technique be applied for general tasks? This question motivates the emergence of another research discipline: neuromorphic computing (NC). In NC, researchers try to identify the most fundamental ingredients that construct intelligence behaviour produced by the brain itself. To achieve this, neuromorphic systems are developed to mimic the brain functionality down to cellular level. In this paper, a neuromorphic platform called SpiNNaker is described and evaluated in order to understand its potential use as a platform for a deep learning approach. This paper is a literature review that contains comparative study on algorithms that have been implemented in SpiNNaker.
Highlights
In recent years, researches on artificial neural networks (ANNs) become a hype again, especially with the rise of learning engines equipped with “deep” learning capability
Deep Neural Network (DNN) is a generalized feedforward neural network with several layers. It is an advanced extension of the first generation of ANN, which was developed by Rosenblatt based on a single thresh-old logic unit neural network proposed by McCulloch and Pitts in 1943 [15]
The complete SpiNNaker system development is underway; when it is complete, it will have 50k chips, in which 1 million ARM cores are available for neuromorphic computing, as well as 7 Tera bytes of memory needed for simulating a massive neural network in biological real-time
Summary
Researches on artificial neural networks (ANNs) become a hype again, especially with the rise of learning engines equipped with “deep” learning capability. Marblestone et al hypothesize that the brain optimizes cost functions, in which the cost functions are diverse and di er across brain locations and over development [12] In this circumstance, the impressive performance demonstrated by DL as a credit assignment through multiple layers of neurons is a manifesto of an optimization operation within a pre-structured architecture matched to the computational problems for specific classes. The impressive performance demonstrated by DL as a credit assignment through multiple layers of neurons is a manifesto of an optimization operation within a pre-structured architecture matched to the computational problems for specific classes Such interpretation does not show a direct relation between learn-ing mechanism in DL and in the brain. DL mechanism does not reflect brain dynamics even though its structure resembles the hierarchical neuronal circuitry of the brain cortex [14]
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
