Deep learning is a kind of machine learning, and machine learning is the only way to realize artificial intelligence.
At present, most of the outstanding AI applications use deep learning technology, leading the third wave of artificial intelligence.
1、 The concept of deep learning
Deep learning is a method of representation learning based on data in machine learning.
It belongs to the category of machine learning. It can be said that it is an upgrade based on the traditional neural network, which is about equal to the neural network. Its advantage is to replace manual feature extraction with unsupervised or semi supervised feature learning and hierarchical feature extraction.
Deep learning is a new field of machine learning. Its motivation is to build and simulate the neural network of human brain for analytical learning. It imitates the mechanism of human brain to interpret data, such as images, sounds and texts.
2、 Advantages and disadvantages of deep learning
The traditional feature extraction of machine learning mainly depends on human. For specific simple tasks, manual feature extraction is simple and effective, but it is not universal. The feature extraction of deep learning does not rely on manual, but is automatically extracted by machine.
The main advantages of deep learning are as follows:
Advantage 1: strong learning ability
From the results, deep learning has a strong learning ability, performance is very good.
Advantage 2: wide coverage and good adaptability
Deep learning neural network has many layers and wide width, and can be mapped to any function in theory, so it can solve very complex problems.
Advantage 3: data driven, high upper limit
Deep learning is highly dependent on data. The larger the amount of data, the better its performance. Especially in the fields of image recognition, face recognition and NLP.
Advantage 4: good portability
Due to the excellent performance of deep learning, many frameworks can be used, such as tensorflow and pytoch. These frameworks can be compatible with many platforms.
Deep learning also has its disadvantages
Disadvantage 1: large amount of calculation and poor portability
Deep learning needs a lot of data and computing power, so the cost is high. And now many applications are not suitable for mobile devices. At present, many companies and teams are developing chips for portable devices.
Disadvantage 2: high hardware requirements
Deep learning requires high computing power, and ordinary CPU can not meet the requirements of deep learning.
Disadvantage 3: complex model design
The model design of deep learning is very complex, which requires a lot of manpower, material resources and time to develop new algorithms and models. Most people can only use ready-made models.
Disadvantage 4: No “human nature”, easy to exist prejudice
Because deep learning relies on data, and its interpretability is not high. In the case of unbalanced training data, there will be gender discrimination, racial discrimination and other issues.
3、 Three typical deep learning algorithms
Convolutional neural network (CNN), recurrent neural network (RNN) and generated countermeasure network (GAN) are three typical deep learning algorithms.
Convolution neural network is a kind of feedforward neural network with depth structure, which contains convolution calculation, and is one of the representative algorithms of deep learning.
CNN has great advantages in image processing, and has been widely used in image classification and retrieval, target location detection, target segmentation, face recognition, bone recognition and other fields.
Cyclic neural network is a kind of recurrent neural network, which takes sequence data as input, recurses in the evolution direction of sequence, and all nodes (cyclic units) are linked in chain.
In the field of deep learning, RNN is an effective algorithm for processing sequence data. It is widely used in text generation, speech recognition, machine translation, generated image description, video tagging and other fields.
As a deep learning model, generative countermeasure network is a popular unsupervised learning algorithm in recent two years.
Gan can generate very realistic photos, images and even videos. It is widely used in many fields, such as generating image data sets, generating face photos, image to image conversion, text to image conversion, image editing, image restoration and so on.