Here I am, exhibiting at my fourth Artomatic event! I love the energy and camaraderie here at Artomatic and I always look forward to these events.
This time, I made these three paintings exclusively for the Artomatic event. I started with three black blank canvasses and decided to apply my “discovery” technique. In the discovery technique, I a sequence of subconscious and deliberate steps to bring out and discover what has been on my mind.
This time the discovery process was particularly hard. I have been doing a very mathematical course in Machine Learning by Geoffrey Hinton while trying to complete these paintings. Two different tasks that require two different parts of the brain. After a number of false starts, the image came clear to me! Of course, it was Hinton’s equations on Backpropagation staring at me all the time. I could not understand why two backs were popping out with such clarity.
Why Deep Learning? Why not just Machine Learning? We will never know, except for the fact that it is a deep question and require deep thought!
The last painting – Training tensorflow was much simpler to understand. The real neural net in the head was trying to learn the virtual neural net by emulating the real neural net in the head. Tensorflow is a Deep Learning package from Google, wherein you simply train the network, a task any kid can do! I perceive the person climbing the learning mountain to be a young boy learning climbing while dreaming Himalayan dreams. Why climbing as against any other activity? My introduction to AI was applying the old Hill Climb Algorithm many-many moons ago.