Neighborhood Component Analysis

Neighborhood Component Analysis (NCA) is a dimensionality reduction technique aiming to learn optimal feature representations that improve the performance of nearest-neighbor classifiers. Current research focuses on enhancing NCA's efficiency and accuracy through modifications like stochastic neighbor sampling and integration with deep learning architectures, leading to improved performance in both classification and regression tasks across various datasets, including those with tabular or graph structures. These advancements demonstrate NCA's continued relevance in machine learning, offering a powerful and efficient approach to feature learning with applications in diverse fields such as biometrics, recommendation systems, and domain adaptation.

Papers

July 3, 2024

Modern Neighborhood Components Analysis: A Deep Tabular Baseline Two Decades Later
Han-Jia Ye, Huai-Hong Yin, De-Chuan Zhan
Deep Model Tabular Deep Learning Tabular Prediction Neighborhood Component Analysis

February 13, 2024

A Novel Approach to Regularising 1NN classifier for Improved Generalization
Aditya Challa, Sravan Danda, Laurent Najman
Novel Approach Simple Classifier Improved Generalization Non Parametric Watershed Algorithm Neighborhood Component Analysis

January 5, 2024

Offline Handwriting Signature Verification: A Transfer Learning and Feature Selection Approach
Fatih Ozyurt, Jafar Majidpour, Tarik A. Rashid, Canan Koc
Transfer Learning Feature Selection Signature Verification Handwritten Signature Neighborhood Component Analysis Offline Signature Offline Handwritten Signature Verification

January 31, 2023

Contrast and Clustering: Learning Neighborhood Pair Representation for Source-free Domain Adaptation
Yuqi Chen, Xiangbin Zhu, Yonggang Li, Yingjian Li, Haojie Fang
Contrastive Learning Unsupervised Domain Adaptation Source Free Domain Adaptation Unlabeled Target Domain Domain Discrepancy Neighborhood Component Analysis

November 25, 2022

Evaluation of the impact of the indiscernibility relation on the fuzzy-rough nearest neighbours algorithm
Henri Bollaert, Chris Cornelis
Global Evaluation Global Impact Rough Set Nearest Neighbour Outcome Indistinguishability Neighborhood Component Analysis

November 15, 2022

Neighborhood Convolutional Network: A New Paradigm of Graph Neural Networks for Node Classification
Jinsong Chen, Boyu Li, Kun He
Graph Neural Network Graph Representation Learning Node Classification Graph Convolutional Network Convolution Layer New Paradigm Neighborhood Component Analysis

September 25, 2022

Flattened Graph Convolutional Networks For Recommendation
Yue Xu, Hao Chen, Zengde Deng, Yuanchen Bei, Feiran Huang
Effective Recommendation Graph Convolutional Network Large Scale Recommendation Quater GCN Neighborhood Component Analysis

January 7, 2022

Neighbor2vec: an efficient and effective method for Graph Embedding
Zhiming Lin
Graph Drawing Link Prediction Effective Method Neighbor Sampling Neighborhood Component Analysis Multi Label Node Classification

December 8, 2021

Revisiting Contrastive Learning through the Lens of Neighborhood Component Analysis: an Integrated Framework
Ching-Yun Ko, Jeet Mohapatra, Sijia Liu, Pin-Yu Chen, Luca Daniel, Lily Weng
Contrastive Learning Representation Learning Unified Framework Camera Lens Self Supervised Representation Learning Neighborhood Component Analysis

Neighborhood Component Analysis

Papers

Modern Neighborhood Components Analysis: A Deep Tabular Baseline Two Decades Later

A Novel Approach to Regularising 1NN classifier for Improved Generalization

Offline Handwriting Signature Verification: A Transfer Learning and Feature Selection Approach

Contrast and Clustering: Learning Neighborhood Pair Representation for Source-free Domain Adaptation

Evaluation of the impact of the indiscernibility relation on the fuzzy-rough nearest neighbours algorithm

Neighborhood Convolutional Network: A New Paradigm of Graph Neural Networks for Node Classification

Flattened Graph Convolutional Networks For Recommendation

Neighbor2vec: an efficient and effective method for Graph Embedding

Revisiting Contrastive Learning through the Lens of Neighborhood Component Analysis: an Integrated Framework