What are the advantages of deep learning compared with traditional machine learning?

The main difference between deep learning and traditional machine learning is that its performance improves with the increase of data scale. When there is little data, the performance of deep learning algorithm is not good. This is because deep learning algorithms need a lot of data to understand perfectly.

Third, hardware dependence.

Deep learning algorithm needs a lot of matrix operations, and GPU is mainly used to optimize matrix operations efficiently, so GPU is the necessary hardware for deep learning to work normally. Compared with traditional machine learning algorithms, deep learning relies more on high-end machines equipped with GPU.

Second, feature processing.

Feature processing is the process of putting domain knowledge into feature extractor to reduce the complexity of data and generate a better model for learning algorithm. Feature processing is time-consuming and requires professional knowledge.

Deep learning attempts to obtain high-level features directly from data, which is the main difference between deep learning and traditional machine learning algorithms. Based on this, deep learning reduces the work of designing feature extractors for each problem.

For example, convolutional neural networks try to learn the underlying features in the previous layer, then learn some faces, and then describe advanced faces. For more information, please read the interesting application of neural network machine in deep learning.

When using traditional machine learning algorithm to solve a problem, traditional machine learning usually decomposes the problem into multiple sub-problems and solves them one by one, and finally combines the results of all sub-problems to get the final result. On the contrary, deep learning advocates direct end-to-end problem solving.