Machine Learning vs Deep Learning: Key Differences | Course Playlist

Key Takeaways

now if you remember in the last video we touched upon this difference already in case of machine learning the data scientist needs to build the features and the more features the data scientist can think of the better is the model whereas in case of deep learning the model itself learns the features from the training data for example example if we provide a deep learning model with images of various faces it will extract several low-level features like boundaries of eyes or nose or ears and then combine them to build midlevel features these mid-level features then combine to build several highlevel features which we can relate to by seeing at these features so this is how deep learning is fundamentally different from machine learning learning the next difference between deep learning and machine learning is that deep learning typically performs better as we provide more data to the model and this is because of the power and the flexibility of deep learning as we provide more data to the model the model will be able to build either better features or more features because of its flexibility which leads to Improvement as we provide more and more data to the model this is not the case with older machine learning algorithms which tend to Plateau after a particular amount of data the next difference between deep learning and machine learning is the amount of computational power required by a deep learning model in fact the requirement is so high that these models are typically run on special Hardware like gpus or tpus if we run these deep learning models on CPUs they will end up taking very large amount of training time to build even the simple models for more differences between gpus and CPUs kindly refer to a video later in the course now because deep learning learns features on its own these models can become difficult to interpret compared to our machine learning models for example a decision tree model is usually easy to understand and even to explain to people but that is not always the case with deep learning this is actually a very active area of research and there are a few ways in which we can interpret deep learning models to some degree but in general machine learning models are easier to interpret compared to deep learning models so in summary these are some of the important differences between deep learning and machine learning well deep learning extract features automatically and usually improves as provide more data it ends up being less interpretable and requires lot more computational power compared to machine learning