Neural Network Orchestration for Dynamic Media Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Businesses face challenges in processing and analyzing big data, particularly due to the vast amount of unstructured and semi-structured data that conventional methods struggle to handle, leading to incomplete insights and inefficient neural network engine selection for specific data types.
Innovation Solution
The Veritone AI platform employs conductor and inter-class technologies to dynamically select the best neural network engines based on input media characteristics, combining layers from various pre-trained neural networks to create a highly accurate and efficient classification system, which can be updated without human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If conventional data warehouse methods are used to process big data, then structured data can be analyzed, but unstructured and semi-structured data cannot be effectively handled
Solution Approach 1:
The patent applies universality by creating a unified neural network orchestration system that can handle multiple data types (structured, unstructured, and semi-structured) through a single platform. The system dynamically selects and orchestrates appropriate neural network engines based on input data characteristics, enabling one system to perform multiple classification functions that previously required separate specialized systems.
2Measurement precision
If multiple neural network engines are used to handle diverse data types, then classification accuracy improves, but system complexity increases
Solution Approach 1:
The patent applies the intermediary principle by introducing an orchestration layer that acts as a mediator between the user and the complex ecosystem of neural network engines. This orchestration system analyzes input data characteristics, automatically selects the most appropriate neural network engines, and coordinates their execution, thereby shielding users from the underlying complexity while maintaining high classification accuracy through optimized engine selection.
Solution Approach 2:
The patent applies dynamics by implementing a dynamic selection mechanism that adapts the neural network engine configuration based on the specific characteristics of the input data. Rather than using a fixed architecture, the system dynamically determines which neural network engines to deploy and how to orchestrate them, allowing the system complexity to scale appropriately with the data complexity while maintaining optimal performance.
3Measurement precision
If businesses manually select neural network engines for specific tasks, then task-specific accuracy can be optimized, but the time and expertise required increases
Solution Approach 1:
The patent applies preliminary action by pre-configuring and pre-training multiple specialized neural network engines for different data types and classification tasks before runtime. The orchestration system has access to a prepared ecosystem of neural network engines that have already been optimized for specific tasks, allowing it to quickly select and deploy the appropriate engine based on the input data characteristics without requiring time-consuming manual configuration or selection at execution time.
Data Source
AI summary
Rather than randomly selecting neural networks to classify a media file, the conductor can determine which neural network engines (from the conductor ecosystem of neural networks) are the best candidates to classify a particular portion/segment of the media file (e.g., audio file, image file, video files). The best candidate neural network engine(s) can depend on the nature of the input media and the characteristics of the neural network engines. In object recognition and identification, certain neural networks can classify vehicles better than others, while another group of neural networks can classify structures better. The conductor can take out the guess work and construct in real-time an inter-class neural network using one or more layers selected from one or more pre-trained neural network, based on attribute(s) of the media file, to classify the media file.