Media Fingerprint Weakbit Prediction for Hash Lookup Accuracy
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Media fingerprinting systems face reduced accuracy and increased search times due to noise and content modifications, as they rely on exact hash look-ups that may miss detecting modified content instances.
Innovation Solution
Non-parametrically computing weakbits in a media fingerprint codeword by determining the likelihood of bit changes due to noise and content modifications, and toggling these weakbits to perform multiple hash look-ups, thereby improving the recall rate and reducing search latency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If exact hash look-up is used for media fingerprint matching, then search speed is fast, but accuracy is reduced due to noise and content modifications
Solution Approach 1:
The system performs preliminary analysis to identify weak bits in the fingerprint codeword before executing the hash lookup. By pre-computing which bits are most susceptible to changes due to noise and modifications, the system prepares a targeted set of bit variations to check, rather than performing exhaustive searches or relying on single exact matches.
Solution Approach 2:
Instead of performing a complete exhaustive search through all possible bit variations, the system applies partial action by focusing only on the most likely weak bits that need toggling. This selective approach checks a limited subset of critical bit positions, providing sufficient accuracy improvement without the full computational cost of examining all possibilities.
2Reliability
If multiple hash look-ups are performed by toggling weakbits, then recall rate is improved, but search latency increases
Solution Approach 1:
The system applies local quality by treating different bits in the fingerprint codeword differently based on their individual characteristics. Instead of uniformly toggling all bits or using a fixed pattern, the system identifies specific weak bits that have higher susceptibility to changes and focuses the multiple look-ups on toggling only those particular bit positions, optimizing the balance between recall improvement and computational overhead.
3Measurement precision
If non-parametric weakbit computation is used to predict bit changes, then matching accuracy is enhanced, but computational complexity increases
Solution Approach 1:
The system employs self-service by using the fingerprint data itself to identify its own weak bits. The non-parametric computation analyzes the characteristics of the specific fingerprint codeword being processed and automatically determines which bits are weakest without requiring external reference data, training sets, or pre-computed statistical models. This makes the complexity inherent to the data rather than requiring additional system resources.
Data Source
AI summary
A value is computed for a feature in an instance of query content and compared to a threshold value. Based on the comparison, first and second bits in a hash value, which is derived from the query content feature, are determined. Conditional probability values are computed for the likelihood that quantized values of the first and the second bits equal corresponding quantized bit values of a target or reference feature value. The conditional probabilities are compared and a relative strength determined for the first and second bits, which directly corresponds to the conditional probability. The bit with the lowest bit strength is selected as the weakbit. The value of the weakbit is toggled to generate a variation of the query hash value. The query may be extended using the query hash value variation.