Abstract:In this paper, we reveal that metric learning would suffer from serious inseparable problem if without informative sample mining. Since the inseparable samples are often mixed with hard samples, current informative sample mining strategies used to deal with inseparable problem may bring up some side-effects, such as instability of objective function, etc. To alleviate this problem, we propose a novel distance metric learning algorithm, named adaptive neighborhood metric learning (ANML). In ANML, we design two thresholds to adaptively identify the inseparable similar and dissimilar samples in the training procedure, thus inseparable sample removing and metric parameter learning are implemented in the same procedure. Due to the non-continuity of the proposed ANML, we develop an ingenious function, named \emph{log-exp mean function} to construct a continuous formulation to surrogate it, which can be efficiently solved by the gradient descent method. Similar to Triplet loss, ANML can be used to learn both the linear and deep embeddings. By analyzing the proposed method, we find it has some interesting properties. For example, when ANML is used to learn the linear embedding, current famous metric learning algorithms such as the large margin nearest neighbor (LMNN) and neighbourhood components analysis (NCA) are the special cases of the proposed ANML by setting the parameters different values. When it is used to learn deep features, the state-of-the-art deep metric learning algorithms such as Triplet loss, Lifted structure loss, and Multi-similarity loss become the special cases of ANML. Furthermore, the \emph{log-exp mean function} proposed in our method gives a new perspective to review the deep metric learning methods such as Prox-NCA and N-pairs loss. At last, promising experimental results demonstrate the effectiveness of the proposed method.

Meta-Neighborhoods

Uncertainty Estimation With Neural Processes for Meta-Continual Learning

Meta-Learning with Neural Tangent Kernels

Referee-Meta-Learning for Fast Adaptation of Locational Fairness

Boosting urban prediction tasks with domain-sharing knowledge via meta-learning

Meta-prediction extends human cortical and subcortical reward learning

Adaptive neighborhood Metric learning

Learning from Multiple Cities: A Meta-Learning Approach for Spatial-Temporal Prediction

Neighborhood-Adaptive Structure Augmented Metric Learning

Meta-learning autoencoders for few-shot prediction

A machine learning framework for neighbor generation in metaheuristic search

Neural Metamorphosis

Meta-Prior: Meta learning for Adaptive Inverse Problem Solvers

Meta-Auxiliary Learning for Adaptive Human Pose Prediction

Efficient Meta Learning Via Minibatch Proximal Update

Adaptive Code Completion with Meta-learning.

Adaptive Nearest Neighbor Machine Translation.

Neighborhood-Adaptive Multi-Cluster Ranking for Deep Metric Learning

MetaMove: On Improving Human Mobility Classification and Prediction via Metalearning

A Meta-Learning Approach to Predicting Performance and Data Requirements

Adaptive Neighborhood Graph Construction for Inference in Multi-Relational Networks