News classification
Contact us
- Add: No. 9, North Fourth Ring Road, Haidian District, Beijing. It mainly includes face recognition, living detection, ID card recognition, bank card recognition, business card recognition, license plate recognition, OCR recognition, and intelligent recognition technology.
- Tel: 13146317170 廖经理
- Fax:
- Email: 398017534@qq.com
What is neural network in AI
What is neural network in AI
What is a neural network?
Here's an example of house prices to explain what neural networks are.
Let's look at a simple neural network (single neuron) first.
Suppose we want to build a model for predicting house prices. Now we have data for 6 houses. The input x is the area of the house, the output y is the house price, and the house price model is y = f(x). This is a linear regression problem because the output is a continuous value.
In the figure above, the red cross denotes the house data points. To build a house price forecasting model is to find a function to fit the house data points. From practical considerations, house prices will not be negative, so we have made corrections to the fitted line. At the bottom of the line is a polyline. When the area is less than a certain value, the price is zero. The blue broken line in the above figure is the housing price forecasting model we set.
The above figure can be generalized into the simplest representation of a neural network model, as shown in the following figure:
"Neuro" means a neuron whose function is to complete the function f(x). This house price prediction model (function) is more common in neural network applications. It is a ReLU (Rectified Linear Unit) function, ie, a modified linear unit. The ReLU function graph is as follows:
Next, take a look at the multi-neuron house price prediction example. In fact, large and complex neural networks consist of many neurons, just like Lego bricks.
The resolution of housing prices is usually not just a characteristic of the house area, such as:
Bedrooms: The size of the house and the number of bedrooms usually determine the family size.
Postal code: Resolution of accessibility, ie walkability
Wealth: Determined school quality with zip code
In the above example, the final neural network model is constructed as follows. The input is x (x1, x2, x3, x4) and the output is y. Here you may be wondering, the family size is determined by the size and bedrooms. How can the postal code and wealth be resolved in the figure below? You can think of the weight of postal code and wealth as very small or 0.
Here's an example of house prices to explain what neural networks are.
Let's look at a simple neural network (single neuron) first.
Suppose we want to build a model for predicting house prices. Now we have data for 6 houses. The input x is the area of the house, the output y is the house price, and the house price model is y = f(x). This is a linear regression problem because the output is a continuous value.
In the figure above, the red cross denotes the house data points. To build a house price forecasting model is to find a function to fit the house data points. From practical considerations, house prices will not be negative, so we have made corrections to the fitted line. At the bottom of the line is a polyline. When the area is less than a certain value, the price is zero. The blue broken line in the above figure is the housing price forecasting model we set.
The above figure can be generalized into the simplest representation of a neural network model, as shown in the following figure:
"Neuro" means a neuron whose function is to complete the function f(x). This house price prediction model (function) is more common in neural network applications. It is a ReLU (Rectified Linear Unit) function, ie, a modified linear unit. The ReLU function graph is as follows:
Next, take a look at the multi-neuron house price prediction example. In fact, large and complex neural networks consist of many neurons, just like Lego bricks.
The resolution of housing prices is usually not just a characteristic of the house area, such as:
Bedrooms: The size of the house and the number of bedrooms usually determine the family size.
Postal code: Resolution of accessibility, ie walkability
Wealth: Determined school quality with zip code
In the above example, the final neural network model is constructed as follows. The input is x (x1, x2, x3, x4) and the output is y. Here you may be wondering, the family size is determined by the size and bedrooms. How can the postal code and wealth be resolved in the figure below? You can think of the weight of postal code and wealth as very small or 0.