Recently got interested in AI, particularly how far it has gone. Now I’m wondering if we are close to having the ultimate AI that closely resembles a human being. I have some background on AI (well sort of) during my university days a long time ago :). I did work on a thesis on fingerprint recognition that requires advanced math skills (e.g. Fourier transforms, etc.) but good thing there is already an IBM PC that time to do that for you. I also met someone doing his Masters thesis using neural networks to train a computer to just check a multiple choice exam. If I remember it right, he is using the back propagation algorithm. Don’t quote me on these, I never did try to understand all these stuff at that time, plus it’s actually advance to me. Fast forward today I see there are already a bunch of algorithms that were created and different ways to train an AI and I would be interested to check them out and see for myself if I can create my own AI that I can train from baby stage to adult stage. I have this opinion that it will take years to train an AI just like how we humans learn. I think in an AI, there are many different areas that a machine learning need to be applied. Ultimately, an AI needs to learn how to learn all these stuff.
So I started this post just to compile my findings on the latest technologies / tools related to AI. So far here are some of them and I will continue to update this as I discover more.
- A cheat sheet listing all there is right now in AI, Neural Networks, Machine Learning, Deep Learning & Big Data. It will give you an idea on what has been done or currently being done.
- A program in Python implementing neural networks. Mind you this is really very technical and requires time to understand. I noticed that Python seems to be the language of choice for machine learning, but I only stumbled upon a few. Have to dig in some more.
- There is also Microsoft Azure Machine Learning which would be interesting to check too.