Mysteries and Questions

My fingers tingle in excitement. My mind is a tangled mess of “what about that essay”, “there’s that assignment”, “when is the deadline again?”, “I should probably look over my notes”, and guilt. Guilt at not being able to multitask and do everything simultaneously. Guilt at wondering if one task is more important than another. Guilt at knowing exactly the order of tasks on the priority list, but refusing to accept it.

As Murphy is exasperatingly always right, luck prevented me from watching the new lecture series I found on deep learning (it has thus far been successful for 6 days). The first lecture too had a wonderful introduction, just as we discussed this week during class about good presentational skills and the importance of having strong “hooks” or enticing introductions. This made it even more challenging to look away while deadlines tugged me away. 

Through research and learning on the side, I have grown acutely aware of how mysterious hidden-layers behave. They work like magic, never revealing their true personalities. They use activation functions to determine whether or not the neuron stays and passes its information (propagates forward through the neural network). However, how can I tell which activation function is the best one to use? What about assigning weights and biases? Why does the neural network work the way it does? 

I hope to answer these burning questions I have been holding for a very long time through the new lecture series I found. I most likely will not understand the first time, nor the second, third, nth. Regardless, I will try again, and again, and again - until I feel satisfied with myself. Right now, I am going to keep my fingers tingling, my curiosity blooming and explore a new conceptual world.