Please briefly describe the advantages of deep learning and traditional machine learning.

What are the advantages of deep learning and traditional machine learning?

Advantages: 1: Strong learning ability.

From training, testing to verification, the performance of deep learning is very good, indicating that learning ability is very strong.

Advantages 2: wide coverage and good adaptability.

Its neural network has many layers and large width, which can be mapped to any function in theory and can solve many complicated problems.

Advantage 3: data-driven, with high upper limit.

Deep learning is highly dependent on data, and the greater the amount of data, the better the performance. Some tasks such as image recognition, face recognition and NLP have even surpassed human performance, and the upper limit can be further increased by adjusting parameters.

Advantage 4: good portability.

Due to the excellent performance of deep learning, there are many frameworks that can be used, such as TensorFlow and Pytorch. These frameworks are compatible with many platforms.

Expand:

Five main differences between machine learning and deep learning:

1. Human intervention

For machine learning systems, humans need to identify and manually code application features according to data types (such as pixel values, shapes and directions), while deep learning systems try to learn these features without additional human intervention. Take the facial recognition program as an example.

This program will first learn to detect and recognize the edges and lines of the face, then the more important parts of the face, and finally the overall appearance of the face. Doing so will involve a lot of data, and with the passage of time and the self-training of the program, the probability of correct answer (that is, accurate face recognition) will gradually increase. This kind of training is carried out by using neural network, which is similar to the way the human brain works and does not need human reprogramming.

2. Hardware

Because the amount of data to be processed is different from the mathematical calculation complexity involved in the algorithm used, the deep learning system needs more powerful hardware than the simple machine learning system. One type of hardware used for deep learning is a graphics processing unit (GPU). Machine learning programs can run on low-end machines that don't have that much computing power.

3. Time

We know that the deep learning system needs a huge data set and involves many parameters and mathematical formulas, so the deep learning system will need a lot of training time. Machine learning may take several seconds to several hours, while deep learning may take several hours to several weeks!

4. Method

Algorithms used in machine learning tend to analyze different parts of data and then combine these parts to obtain results or solutions. The deep learning system solved the whole problem at once.

For example, if you want to use a program to identify specific objects in an image (what they are and where they are-such as the license plate of a car in a parking lot), you must complete two steps through machine learning: first, object detection, and then object recognition. With the deep learning program, you only need to input an image, and through training, the program will return the recognized object and its position in the image in a result.

5. Application

Through the above differences, you may have realized that machine learning and deep learning systems will be used in different applications. Usage: Basic machine learning applications include forecasting programs (such as forecasting the price of the stock market or the time and place of the next hurricane), spam identifiers and programs for designing evidence-based treatment plans for medical patients.

In addition to the above-mentioned examples of Netflix, music streaming service and facial recognition, another well-known application field of deep learning is self-driving cars-this program uses multi-layer neural networks to do things, such as determining objects to avoid, recognizing traffic lights and knowing when to accelerate or decelerate.