Active Nearest-Neighbor Learning in Metric Spaces

Authors: Aryeh Kontorovich, Sivan Sabato, Ruth Urner

NeurIPS 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Theoretical We propose a pool-based non-parametric active learning algorithm for general metric spaces, called MArgin Regularized Metric Active Nearest Neighbor (MARMANN), which outputs a nearest-neighbor classifier. We give prediction error guarantees that depend on the noisy-margin properties of the input sample, and are competitive with those obtained by previously proposed passive learners. We prove that the label complexity of MARMANN is significantly lower than that of any passive learner with similar error guarantees.
Researcher Affiliation Academia Aryeh Kontorovich Department of Computer Science Ben-Gurion University of the Negev Beer Sheva 8499000, Israel Sivan Sabato Department of Computer Science Ben-Gurion University of the Negev Beer Sheva 8499000, Israel Ruth Urner Max Planck Institute for Intelligent Systems Department for Empirical Inference Tübingen 72076, Germany
Pseudocode Yes Algorithm 1 Generate NNSet(t, I, δ)
Open Source Code No The paper does not contain any statement or link indicating that the source code for MARMANN is open-sourced or available.
Open Datasets No The paper describes a theoretical framework and does not use or provide access information for any specific real-world or simulated dataset for empirical training. It refers to 'Sin Dm' as a theoretical sample.
Dataset Splits No The paper is theoretical and does not conduct empirical experiments with validation datasets or specific splits.
Hardware Specification No The paper does not mention any specific hardware used, as it focuses on theoretical analysis rather than empirical experimentation.
Software Dependencies No The paper does not mention any specific software dependencies or versions, as it presents theoretical work without empirical implementation details.
Experiment Setup No The paper does not describe any experimental setup details such as hyperparameters or training configurations, as it is a theoretical paper.