Philosophy of Life - Inspired by fast.ai

Are not our lives trained like the fast.ai neural nets? Our parents upbringing built the base model which we try to freeze forever. We are then thrown into the outside world, where we use the base model to lead a normal life. As new situations arise, we are exposed to new datasets. We unfreeze a few layers and retrain our model. A lower learning rate takes a lot of time but builds a stronger us; whereas a higher learning rate might not help us understand what we are actually doing. So its important to determine the optimal learning rate for every situation and every person, as no two nets are the same.

There are times when we are mind-washed by others aren't there? That is when a lot of layers are unfrozen and trained with a higher learning rate!

In general, do we not feel that when we encounter a situation repeatedly, we become an expert in handling it? It is just that we have increased the epochs and trained our model over and over with the same data and boom! we  have a very accurate model. When the neural-net sees the same image multiple times, it reaches a level of saturation and starts deteriorating in performance. Yes! that is the reason why we humans keep looking for change in life. We are an active neural net, constantly looking for new-datasets and new-experiences. So let us keep learning and live active to build a better us!

What we are taught as a child is always hard to erase and that is why growing up in a good environment and learning a lot while we are a kid is always vital!

#fast.ai #JeremyHoward