Neural network pruning method and apparatus
A neural network and pruning technology, applied in the computer field, can solve the problems of compression and acceleration accuracy at the same time, and achieve the effect of small loss of accuracy, good compression and acceleration effects
Inactive Publication Date: 2017-03-29
BEIJING TUSEN ZHITU TECH CO LTD
View PDF0 Cites 36 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0004] In view of the above problems, the present invention provides a neural network pruning method and device to solve the technical problem in the prior art that compression, acceleration, and accuracy cannot be taken into account at the same time
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0032] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described The embodiments are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other embodiments obtained by persons of ordinary skill in the art without making creative efforts shall fall within the protection scope of the present invention.
[0033] The above is the core idea of the present invention. In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, and to make the above-mentioned purposes, features and advantages of the embodiments of the present invention more obvious and understandable, ...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention discloses a neural network pruning method and apparatus, for solving the technical problem of incapability of taking compression, acceleration and precision into consideration in network pruning in the prior art. The method comprises the following steps: according to activation values of neurons in a network layer to be pruned, determining importance values of the neurons; according to connection weights between the neurons in the network layer to be pruned and neurons in a next network layer, determining diversity values of the neurons; according to the importance values of the neurons in the network layer to be pruned and the diversity values, selecting reserved neurons from the network layer to be pruned by use of a volume maximization neuron selection strategy; and obtaining a pruned network layer by cutting other neurons in the network layer to be pruned. According to the technical scheme, a quite good compression and acceleration effect can be achieved while the precision of a neural network is guaranteed.
Description
technical field [0001] The invention relates to the field of computers, in particular to a neural network pruning method and device. Background technique [0002] At present, deep neural networks have achieved great success in the field of computer vision, such as image classification, object detection, image segmentation, etc. However, deep neural networks with good effects often have a large number of model parameters, which not only requires a large amount of calculation, but also occupies a large part of the space in actual deployment, which cannot be used normally in some application scenarios that require real-time computing. Therefore, how to compress and accelerate deep neural networks is particularly important, especially in some future application scenarios that require deep neural networks to be applied to embedded devices and integrated hardware devices. [0003] At present, the way of compressing and accelerating deep neural networks is mainly realized by netwo...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06N3/08
CPCG06N3/082G06N3/048G06F17/16G06N3/04
Inventor 王乃岩
Owner BEIJING TUSEN ZHITU TECH CO LTD