Deep Learning is a word used often these days; though many people don’t fully understand what it truly means. This article aims to help you understand and comprehend what it is.
To dive into Deep Learning, you must understand that it is a subset of Neural Networks, which is a subset of Machine Learning which in turn is a subset of Artificial Intelligence. I know that’s a lot to take in so here’s a picture:
So, it is pretty evident that Deep Learning functions mainly due to Neural Networks.
Now, why has Deep Learning become so famous? To put it in a line — because it is very powerful. Deep learning has surpassed all previously known landmarks set by algorithms in the fields of image recognition, voice recognition, feature extraction, etc.
So, what is Deep Learning exactly?
Deep Learning is a supervised, unsupervised, or semi-supervised technique through which machines learn solely with the help of Neural Networks.
I won’t lie to you, I’m not an expert myself, so when I say ‘solely’, it is based on my knowledge and I haven’t really found any proof otherwise.
How are Neural Networks a part of Deep Learning?
Neural Networks are the basis if Deep Learning, we get the inputs through the input layer of the Neural Network, extract all the features in the hidden layers, and get the result from the output layer. Features are basic characteristics that help the model associate the input to the output.
So, if I were to pass a picture of a dog as the input, the hidden layers may extract features such as droopy ears or some other feature and associate it with the dog. If I were to pass a picture of a cat however as the input, the hidden layers may extract features such as pointed ears and associate it with the cat.
To sum it up, Deep Learning relies heavily on Neural Networks, and therefore it is able to outperform Machine Learning and hasn’t hit the “upper limit” yet. Many of you may wonder what is the “upper limit”, so as you all know the more data you have to work with the better. The precision and accuracy of a model can be increased by feeding it more data, but there comes a point where the model plateaus i.e. feeding it more data renders useless. This is the “upper limit”, which hasn’t been reached while working with Deep Learning models.
If you have queries feel free to tag us down below and I’ll answer.
If you want us to start a series on Neural Networks from basic let us know.
