New algorithms offer exact Data Shapley for weighted KNN regression

Researchers have developed new algorithms for calculating Data Shapley values in weighted k-nearest-neighbor (KNN) regression and soft-label prediction. Previously, exact calculations for these scenarios were computationally infeasible, relying on brute-force methods that were exponential in complexity. The new methods include a pseudo-polynomial-time exact algorithm and a certified fully polynomial-time approximation scheme (FPTAS) for continuous weights, offering deterministic results and certified error bounds. These advancements are crucial for accurately assessing the value of training data points in complex machine learning models and have been released as an open-source library. AI

IMPACT Provides exact and certified methods for data valuation, improving the interpretability and auditing of machine learning models.

RANK_REASON Academic paper detailing new algorithms for data valuation in machine learning. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.AI →

• Data Shapley
• Fully polynomial-time approximation scheme
• k-nearest-neighbor (KNN)
• Monte-Carlo Data Shapley
• OpenDataVal
• pyDVL

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →