I was a bit distracted when listening to the video lectures of Coursera Machine Learning week 4. Fortunately it was a short session, so hopefully I didn’t miss much. This week’s lectures were about neural networks. Neural networks are networks consisting of neurons. Not real neurons, but computer simulations thereof.
Neural networks are useful for non-linear problems such as recognizing objects in pictures. In this module we also get a short tutorial on neurons in the brain. As far as I understand Machine Learning only uses crude approximations of real neurons. Neurons in Machine Learning receive numerical inputs and create numerical outputs. Output of one neuron can serve as input for another. Neurons are typically organized in layers. A neural networks has several layers.
Armchair Hedge Fund
It seems to me that neural networks are good at recognizing patterns. Pattern recognition is very important in trading. For instance, we have the famous “Head and Shoulders” pattern. Can a neural network recognize this pattern? I think so. Or maybe it’s better at recognizing the Golden Cross with Cup Handle. Inverted or otherwise. Personally I would be happy if the neural network could recognize a breakout. That would be enough. Apparently the ann neural networks module has been removed from scikit learn and instead you are redirected to pybrain. I wonder why that is. Anyway I think I have figured out what my homework for this week will be.