Tue Jun 18 2013, 15:10
hitesh technoinspiron Research claim that its Deep Neural Network (DNN) technology has helped improve the performance of Microsoft's Bing voice search.
Microsoft Research is one of the leading technology research outfits and works on a wide variety of problems. The firm has been working on neural networks for several years and work presented last year has been incorporated into Bing voice search, with Microsoft claiming it has improved latency and voice recognition capabilities of the service.
According to Microsoft, its DNN provides a pattern recognition service. While neural networks have been around for decades, the firm has been refining the algorithms that its DNN runs and claims its DNN has significantly improved voice recognition performance.
Dong Yu, senior researcher at Microsoft Research's Conversational Systems Research Center said, "Our result significantly advanced the state of the art, both in industry and in the academic community. Now, most industrial automatic-speech-recognition systems are DNN-based. This also helped to popularize deep learning."
Yu said that DNN could be used for more than just voice recognition. "Before our result, deep learning was only tested on small tasks and did not attract wide attention. I believe this is just the first step in advancing the state of the art. Many difficult problems may be attacked under this framework, which might lead to even greater advances," he said.
Microsoft's neura might be most visible in tasks such as voice recogition but pattern matching is particularly useful in analysing large quantities of data for imformation that can and this is where Microsoft is likely to make some money from its research
.webp)
.webp)