Before learning the deep learning frameworks such as tensorflow and caffe, you need to understand some basic concepts. This article records the basic concepts of a zero-based white that needs to be understood first.
The relationship between artificial intelligence, machine learning and deep learning
ArTIficial Intelligence - the intelligence that gives the machine
"General AI": A omnipotent machine that has all our perceptions (even more than people), all our rationality, can think like us
Narrow AI: Weak artificial intelligence is a technology that can perform a specific task, just like people, even better than people. For example, image classification on Pinterest; or Facebook face recognition.
Strong artificial intelligence is the vision, and weak artificial intelligence is currently achievable.
Machine learning - a way to implement artificial intelligence
The most basic approach to machine learning is to use algorithms to parse data, learn from it, and then make decisions and predictions about events in the real world.
Deep learning - a technology for machine learning
Machine learning can be achieved through neural networks. Deep learning can be simply understood as a machine learning method that uses a deep architecture such as a deep neural network. Most of the current deep architecture refers to deep neural networks.
Neural network composition
A neural network consists of many neurons, each circle being a neuron, and each line represents the connection between neurons. The input data represented by x, y represents the output data, and w represents the weight of each layer connection. w is also what we need to determine after we have constructed the neural network.
The leftmost one is called the input layer, which is responsible for accepting input data.
The far right is called the output layer, we can get the neural network output data from this layer.
The input layer and the output layer are called hidden layers. The number of hidden layers is variable. A simple neural network may be 2-3 layers, and complex ones may be hundreds of layers. A hidden layer is called a deep neural network.
Deep networks are more expressive than shallow networks and can handle more data. But the training of deep networks is more complicated. Need a lot of data, a lot of skills to train a deep network.
Question: Assuming that the calculation speed is fast enough, is the depth of the network as deep as possible?
No. The deeper the deep network, the higher the requirements for architecture and algorithms. After exceeding the bottleneck of architecture and algorithms, it is futile to increase the depth.
Neuron (perceptor)The neural network consists of individual neurons, and one neuron is composed of three parts.
Input weights Each input will have a weight w and an offset value b. That is w0 in the figure. The process of training a neural network is actually the process of determining the weight w.
After the activation function passes the weight operation, it will also go through the activation function and output. For example, we can use the step function f to represent the activation function.
Output the final output, the output of the sensor can be represented by this formula
Neurons can fit arbitrary linear functions, such as the simplest fit and function.
The and function truth table is shown above. Take w1 = 0.5; w2 = 0.5 b = -0.8. The activation function takes the step function f representation of the above example. It can be verified that the neuron can represent the and function at this time.
If you enter the first line, x1 = 0, x2 = 0, you can get
y is 0, which is the first line of the truth table.
In mathematical terms, you can understand the neurons of the and function. It represents a linear classification problem, which is like a straight line separating the classification 0 (false, red cross) and the classification 1 (true, green dot).
In fact, neurons can be understood mathematically as a data segmentation problem. Neurons are the key to transforming neural networks into mathematical problems. For example, if you need to train a neural network to be a classifier, you can mathematically understand the input parameters (x1, x2..., xn) as n points on the m-dimensional coordinate system (where x is an m-ary vector). Each neuron can be understood as a fitting function. Take m to 2 and put it in the simplest two-dimensional coordinate system for understanding.
We have experience and skill to support customers to tooling for their required waterproof connectors, like IP68 series,micro fit connectors. Etop wire assemblies for various industries have been highly recognized by all the customers and widely used for automobiles, electrical and mechanical, medical industry and electrical equipemnts, etc. Products like, wire harness for car audio, power seat, rear-view mirror, POS ATM, Diesel valve Cover gasket fit, elevator, game machine, medical equipment, computer, etc.
JST Connector,Molex Connector, Multi-Contact Connector, Micro Fit Connectors
ETOP WIREHARNESS LIMITED , https://www.wireharnessetop.com