Instance Level Loss

Instance-level loss functions are emerging as a powerful tool in machine learning, aiming to improve model performance by assigning different weights or penalties to individual data points based on their characteristics or difficulty. Current research focuses on adapting these losses within various contexts, including knowledge distillation, aggregate learning, and robust training against adversarial attacks, often employing techniques like adaptive weighting schemes and Gaussian kernel-based similarity measures. This approach offers significant advantages in scenarios with imbalanced data, privacy constraints, or the need for enhanced robustness, leading to improved accuracy and generalization in diverse applications such as acoustic scene classification and cross-modal retrieval.

Papers

November 15, 2024

Forming Auxiliary High-confident Instance-level Loss to Promote Learning from Label Proportions
Tianhao Ma, Han Chen, Juncheng Hu, Yungang Zhu, Ximing Li
Planning Loss Label Proportion Enhanced Learning Instance Level Loss

October 21, 2024

Improving Instance Optimization in Deformable Image Registration with Gradient Projection
Yi Zhang, Yidong Zhao, Qian Tao
Optimization Purpose Deformable Image Registration 3D Brain Gradient Projection Instance Optimal Instance Level Loss

May 11, 2024

AdaKD: Dynamic Knowledge Distillation of ASR models using Adaptive Loss Weighting
Shreyan Ganguly, Roshan Nayak, Rakshith Rao, Ujan Deb, Prathosh AP
Knowledge Distillation ASR Model Adaptive Loss Knowledge Distillation Loss Adaptive Knowledge Distillation Instance Level Loss Conventional Knowledge Distillation

January 20, 2024

Learning from Aggregate responses: Instance Level versus Bag Level Loss Functions
Adel Javanmard, Lin Chen, Vahab Mirrokni, Ashwinkumar Badanidiyuru, Gang Fu
LeArning Abstract Loss Function Planning Loss Multiple Instance Private Prediction Instance Level Loss Aggregate Response

November 6, 2023

DRAUC: An Instance-wise Distributionally Robust AUC Optimization Framework
Siran Dai, Qianqian Xu, Zhiyong Yang, Xiaochun Cao, Qingming Huang
Distributionally Robust Optimization Area Under the ROC Curve Instance Level Loss

November 1, 2023

Enhancing Clustering Representations with Positive Proximity and Cluster Dispersion Learning
Abhishek Kumar, Dong-Gyu Lee
Deep Clustering Clustering Friendly Representation Robust Clustering Close Proximity Learning Based Point Clustering Benchmark Instance Level Loss

May 7, 2023

Instance-Variant Loss with Gaussian RBF Kernel for 3D Cross-modal Retriveal
Zhitao Liu, Zengyu Liu, Jiwei Wei, Guan Wang, Zhenjiang Du, Ning Xie, Heng Tao Shen
Cross Modal Cross Modal Retrieval Instance Level Loss

March 24, 2023

Improved Adversarial Training Through Adaptive Instance-wise Loss Smoothing
Lin Li, Michael Spratling
Adversarial Attack Adversarial Training Adversarial Robustness Adversarial Vulnerability Instance Level Loss

March 16, 2022

Instance-level loss based multiple-instance learning framework for acoustic scene classification
Won-Gook Choi, Joon-Hyuk Chang, Jae-Mo Yang, Han-Gil Moon
New Framework Multiple Instance Learning Convolutional Filter Acoustic Scene Classification Multiple Instance Acoustic Scene Instance Level Label Instance Level Loss

Instance Level Loss

Papers

Forming Auxiliary High-confident Instance-level Loss to Promote Learning from Label Proportions

Improving Instance Optimization in Deformable Image Registration with Gradient Projection

AdaKD: Dynamic Knowledge Distillation of ASR models using Adaptive Loss Weighting

Learning from Aggregate responses: Instance Level versus Bag Level Loss Functions

DRAUC: An Instance-wise Distributionally Robust AUC Optimization Framework

Enhancing Clustering Representations with Positive Proximity and Cluster Dispersion Learning

Instance-Variant Loss with Gaussian RBF Kernel for 3D Cross-modal Retriveal

Improved Adversarial Training Through Adaptive Instance-wise Loss Smoothing

Instance-level loss based multiple-instance learning framework for acoustic scene classification