DailyPost 142
DEEP LEARNING TECHNOLOGY
When human beings are getting mesmerised by shallow learning; deep learning technology is making major strides with a potential of changing our lives completely in the days to come. Siri, DeepMind / AlphaGo et al are already giving us the feel of the world to come. At the core of deep learning is the use of artificial neural networks. In simple terms these are programs which make the machine to learn.
Based on a set algorithms, a branch of machine learning, intends to model high level abstractions of data. Artificial neural networks are algorithms that enable the machine to learn, inspired by the computation in our brains. Artificial neural networks are connected as the real neurons. It detects patterns. Deep artificial neural networks have layers. It was a lost cause ten years ago, today it had become a rage.
Lots of research papers were being published but no great progress was made in real terms. The deep learning community was growing. Computational power had become a big challenge. Graphics card was the solution and became the way forward. Computational power of a super computer was available for a few thousand dollars and the results started coming. The steady increase in computational power helped deep learning considerably. Speech recognition took a big leap because of deep learning.
Libraries grew up to support the deep learning community, open source & robust code libraries. It’s quite easy now to get started. It’s like playing with Legos. Deep learning was in the lead of computer vision.
New reviewing model of research work came into existence. The work is put on the website and discussions start straightaway. Open social media discussions take place much before the official validation. Graphics Card and Open Source Library support facilitates faster and intense research, making it vastly iterative and thus progressive – real time and dynamic.
PATH BREAKING SPEECH RECOGNITION AND COMPUTER VISION IS FUELLED BY DEEP LEARNING.
Sanjay Sahay