Before starting the #100daysofMLcode challenge, I was quite confused between the difference of machine learning and deep learning. So, lets discuss the workflow of ML and DL.
In Machine learning, first we collect training data, then we extract features from the training data then we build a model( an algorithm) to predict the output value.
- It can even work on small dataset. Increasing data will improve accuracy.
- Training time is lower compared to deep learning.
- It doesn’t need high performance hardware.
Deep learning is the sub class of machine learning. It deals with complex model which can’t be solved by the traditional machine learning technique. From traditional machine learning technique, I meant that we can’t extract feature from a large amount of data. That’s why we use deep learning technique.
In this we don’t need to extract features like machine learning. We just have to put our training data in the deep learning model which contains multiple deep layers of neural network for feature extraction. Thus the model will learn the features from the data and thus we can get predict output for unseen values.
- No algorithm is required
- No need of feature extraction, DL model will do that for you.
- Large amount of data is required.
- Training time is higher compared to machine learning.
- Good hardware is required to run model.