Temporal Neural Network Configuration Switching for Natural Video
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Neural networks exhibit variations in output due to their structural characteristics, leading to unnatural video quality when processing temporally consecutive images, as predicted values differ across node connections where error back propagation is not performed.
Innovation Solution
Implement a neural network configuration switching mechanism that alternates between different input-output temporal relationships and node connections during training, using a determining unit to switch configurations based on evaluation of image quality variation, and training with error back propagation to minimize adverse effects on image quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a neural network with fixed node connections is used for processing temporally consecutive images, then the network structure is simple and easy to implement, but the output quality varies across different node connections leading to unnatural video appearance
Solution Approach 1:
The patent applies the dynamics principle by making the neural network configuration changeable during training. The determining unit dynamically selects between first and second configurations based on input data characteristics, and the configuration changing unit adjusts node connections and input-output temporal relationships in real-time. This dynamic adaptation allows the network to maintain consistent output quality across different node connections while processing temporally consecutive images, resolving the contradiction between quality consistency and structural simplicity.
Solution Approach 2:
The patent implements parameter changes by modifying the neural network's structural parameters (node connections and input-output temporal relationships) based on the determined configuration. The configuration changing unit changes these parameters dynamically during training to switch between different configurations, enabling the network to adapt to different processing scenarios and maintain consistent output quality without requiring a permanently complex structure.
2Reliability
If multiple neural network configurations are used to reduce output variation, then the image quality consistency improves, but the training complexity and computational overhead increase
Solution Approach 1:
The patent uses dynamics by implementing a determining unit that dynamically selects the appropriate configuration during training based on input characteristics. This dynamic selection mechanism allows the system to leverage multiple configurations for improved reliability while managing training complexity through intelligent, data-driven configuration selection rather than exhaustive training of all configurations simultaneously.
Solution Approach 2:
The patent applies self-service through the determining unit and configuration changing unit that automatically select and adjust neural network configurations based on input data characteristics without requiring manual intervention. This self-service mechanism reduces training system complexity by automating the configuration management process, allowing the system to achieve improved output consistency through autonomous adaptation.
Data Source
Figure 1
Figure 2A~2B
Figure 3
AI summary
An information processing apparatus includes: an inference unit to which a neural network is applied, the neural network being configured to use a set of temporally consecutive images and a set of temporally consecutive image features as inputs, and output a set of images and a set of image features for the inputs; a determining unit configured to determine which one of first and second configurations to which the neural network is to be configured, the first configuration having a first input-output temporal relationship and the second configuration having a second input-output temporal relationship different from the first input-output temporal relationship; a configuration changing unit configured to change a node connection and an input-output temporal relationship of the neural network to change the configuration of the neural network according to the result determined by the determining unit; and a training unit configured to train the neural network.