Classifier Training Tempered Indication for Overprediction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Prior art systems tend to overpredict the probability of an entity being in a particular class when there are large numbers of low weight indicators in social media publications, leading to inaccurate classification.
Innovation Solution
A system for training a classifier that uses a tempered indication, calculated as a ratio of a weighted sum of tokens divided by a tempering factor, which is a function of the unweighted sum of tokens, to reduce overprediction by mitigating the impact of low weight indicators.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If prior art systems assign different weights to different indicators, then classification accuracy is improved for strong indicators, but overprediction occurs when large numbers of low weight indicators are present
Solution Approach 1:
The patent changes the parameter calculation by introducing a tempering factor that transforms the weighted sum of indicator values into a tempered indication. This parameter transformation ensures that the contribution of low-weight indicators is appropriately limited, preventing overprediction while maintaining sensitivity to strong indicators. The tempered indication is calculated as the weighted sum divided by a tempering factor that increases with the number of indicators present.
2Adaptability or versatility
If the system considers more indicators, then comprehensive classification coverage is improved, but the impact of low weight tokens causes overprediction
Solution Approach 1:
The tempering factor acts as an intermediary between the raw indicator values and the final classification probability. It mediates the influence of multiple low-weight indicators by scaling down their collective impact, thereby preventing overprediction while still allowing the system to consider a comprehensive set of indicators for classification.
3Reliability
If weighted sums of tokens are used, then strong indicators have greater impact, but large numbers of low weight tokens accumulate to cause overprediction
Solution Approach 1:
The tempering factor serves as a counterweight that offsets the cumulative effect of numerous low-weight indicators. By dividing the weighted sum by this factor, the system counterbalances the tendency of multiple small contributions to accumulate and cause overprediction, while preserving the differential impact of strong indicators.
Data Source
AI summary
An artificial intelligence system for training a classifier has a database of training data and a modeling system for building a classification model based on the training data. The database has a binary classification for each entity and binary tokens indicating whether or not one or more indicators about the entity are true. The classification model is based on a tempered indication of the tokens. The tempered indication is a ratio of a weighted sum of the tokens for each entity divided by a tempering factor for each of the entities. The tempering factor is a function of the unweighted sum of the tokens for each entity. Thus the tempering factor will reduce the tempered indication when large numbers of low weight tokens are present so that the model does not over predict the probability of an entity being in the classification.