The goal of cognitive computing is to simulate human thought processes in a computerised model. Using self-learning algorithms that use data mining, pattern recognition and natural language processing, the computer can mimic the way the human brain works.
However, Raghavendra Singh from IBM Research, while speaking at Cypher 2017, added that no cognition is possible without a level of interaction.
Speaking about computer vision being the main area for deep learning, Singh gives insights on how techniques take a connectionist approach to deep structure learning. Additionally, he goes on to talk about and explain how neural networks work.
Human vs computer vision
Sighting an example, Singh explains how it is easy for the human mind to see and identify a cat; but it turns out that trying to get a computer to figure that out is an extremely difficult task. However, the herculean task of developing cognitive computing continues. We must all be aware of the ImageNet project designed for use in visual object recognition software research. As of 2016, over ten million URLs of images have been hand-annotated by ImageNet to indicate what objects are pictured; in at least one million of the images, bounding boxes are also provided.
Convolutional neural networks in computer vision
Convolutional neural networks have been some of the most influential innovations in the field of computer vision. We know how CNNs take a biological inspiration from the visual cortex. The visual cortex has small regions of cells that are sensitive to specific regions of the visual field, Singh goes on to explain this in greater detail in his talk.
He also makes a reference to how CNN was expanded upon by a fascinating experiment by Hubel and Wiesel in 1962 (Video) where they showed that some individual neuronal cells in the brain responded (or fired) only in the presence of edges of a certain orientation.
How e-commerce uses cognition?
There has been much talk about the need for creating engagement among the customers on e-commerce sites. Singh points out how in the presence of an assistant at a brick and mortar shop, buyers may feel more at ease. But how are e-commerce companies bridging the gap?
A pair of new image-analysis solutions could open the door for powerful e-commerce and retail applications that significantly improve shopping experiences or create new marketing opportunities.
The advantages of visual product discovery are many folds. It can surface a greater proportion of a product catalogue and find products in separate categories that a customer may not have otherwise encountered
But there are more challenges that we know or can think of. Raghavendra Singh does an in-depth analysis of the challenges of computer vision. Watch the full video for deeper insights.
Try deep learning using MATLAB