High Similarity
High similarity research focuses on developing methods to effectively measure and leverage similarities between data points, whether they are images, text, or neural network representations. Current research emphasizes the use of transformer models, graph neural networks, and various similarity metrics (e.g., cosine similarity, embedding similarity) to achieve this, often within the context of specific applications like image retrieval, anomaly detection, and multi-task learning. This work is significant because improved similarity assessment enhances the efficiency and accuracy of numerous machine learning tasks, impacting fields ranging from computer vision and natural language processing to copyright protection and personalized recommendations.
Papers
Treasure What You Have: Exploiting Similarity in Deep Neural Networks for Efficient Video Processing
Hadjer Benmeziane, Halima Bouzidi, Hamza Ouarnoughi, Ozcan Ozturk, Smail Niar
Similarity of Neural Network Models: A Survey of Functional and Representational Measures
Max Klabunde, Tobias Schumacher, Markus Strohmaier, Florian Lemmerich