What does it do?
According to Mitsubishi, the algorithm adapts to the specific purposes of each system and designs deep learning automatically without any expert developers knowledge and this is done by using learning data and high-level inferences to each instrument’s environment based on learning data. The company expects this to support the effective structuring of networks and reduce the trial and error of design.
How will it help?
The company states that the algorithm will shorten development time and reduce costs compared to current methods in which professionals and developers manually design AI. Mitsubishi claims that the development of AI using this algorithm is expected to require from just a few minutes to a few hours, whereas, development by professionals can take from a few days to a few of weeks. Hence, it will eliminate the need for specialized knowledge or experts, thereby lowering the cost of development.
How does it work?
The algorithm adapts to the specific usage environment, which the company says is essential for an effective AI. The new Automated Design Deep Learning Algorithm designs the structures and initial parameters of deep learning by extracting the most characteristic data from learning data without overlaps. Conventionally, this required the involvement of experts making high-level inferences about specific environments. The idea is based on “Neocognitron,” image processing algorithm considered based on visual cortex.
Quoting an E&Y report, Mitsubishi states that the AI market was estimated to be worth $35 billion (approx. 3.6 trillion yen) in 2015 and annual growth is expected to average 30%. The new system will enable AI to be used in diverse business fields, such as high-level information processing.
Mitsubishi Electric will be presenting this new system at the International Conference on Neural Information Processing from October 16th through 21st at Kyoto University and will publish a paper on the same.