Sound And Visual Signal Segmentation for Adaptive Neural Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for processing sound or visual content using artificial neural networks often require significant computational resources and are not optimized for adaptability to different sub-parts of the content, leading to suboptimal quality in decoded signals.
Innovation Solution
Distributing samples of the signal into multiple sets and using a different, simplified artificial neural network for each set, tailored to its specific characteristics, with parameters optimized to minimize distance from the original signal, allowing for improved reconstruction quality without excessive computational overhead.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a single complex artificial neural network is used to process the entire signal, then the processing quality may be improved, but the computational resources and complexity increase significantly
Solution Approach 1:
The signal is divided into multiple subsets, and a separate simplified neural network is trained and applied to each subset. This segmentation allows the use of less complex networks for each portion while collectively achieving good reconstruction quality, resolving the contradiction between quality and complexity.
Solution Approach 2:
Different simplified neural networks are applied to different subsets of the signal, allowing each local region to be processed by a network optimized for its specific characteristics. This local approach maintains overall quality while keeping individual network complexities low.
2Manufacturing precision
If a single complex artificial neural network is used to process the entire signal, then the processing quality may be improved, but the computational resources required increase
Solution Approach 1:
By segmenting the signal into subsets and processing each with a simplified network, the total computational load is distributed and reduced compared to running one large complex network on the entire signal, while maintaining reconstruction quality.
Solution Approach 2:
Instead of applying one complex network to all signal data, the approach uses multiple partial applications of simpler networks to subsets of the data, achieving comparable or sufficient quality with reduced overall computational resource consumption.
3Adaptability or versatility
If different artificial neural networks are used for different sub-parts of the signal, then the adaptability to specific sub-parts is improved, but the device complexity increases
Solution Approach 1:
The signal is segmented into subsets, allowing different simplified neural networks to be applied to each subset. This segmentation enables adaptability to local signal characteristics while keeping the number of networks manageable by dividing the overall task into smaller portions.
Data Source
Figure 1~2
Figure 3~4
AI summary
A method for processing values associated with samples of a signal representative of sound or visual content, comprises the following steps: - distribution (E20) of the samples within a plurality of sets; - for at least some of the sets of the plurality of sets, decoding (E22) of data (NNCi) representative of parameters defining an artificial neural network associated with the set concerned and processing (E24) of the values associated with the samples of the set concerned by means of the artificial neural network defined by said parameters so as to produce processed values respectively associated with the samples of the set concerned. Also described are an associated device, as well as a method and a device for generating parameters used for processing the aforementioned values.