Algorithm and Data Imbalance: Implementing the New IIL Benchmarks
7 Nov 2025
This section provides an algorithm overview for the proposed Instance-Incremental Learning (IIL) method and details the creation of its benchmarks
KC-EMA Mechanism: Theoretical Analysis and Derivation for IIL
7 Nov 2025
This section provides the detailed theoretical derivation for the KC-EMA mechanism in Instance-Incremental Learning
Future of IIL: Narrowing the Gap and Advancing Knowledge Accumulation
7 Nov 2025
This conclusion validates a novel IIL setting that aims for cost-effective model enhancement using only new data.
Ablation: The Role of Fused Labels and Teacher EMA in Instance-Incremental Learning
7 Nov 2025
This ablation study on Cifar-100 validates the three core components of the proposed IIL method.
SAGE Net Ablation Study: Analyzing the Impact of Input Sequence Length on Performance
5 Nov 2025
This article compares the proposed IIL method against SOTA incremental learning techniques on Cifar-100 and ImageNet-100.
Evaluating Instance-Incremental Learning: CIL Methods on Cifar-100 and ImageNet
5 Nov 2025
This section outlines the experimental setup for the new Instance-Incremental Learning benchmarks.
Model Promotion: Using EMA to Balance Learning and Forgetting in IIL
5 Nov 2025
This article introduces a novel Knowledge Consolidation strategy for IIL that utilizes Exponential Moving Average to transfer learned knowledge
A Graph Transformer Network for Predicting Remaining Process Time
5 Nov 2025
When it comes to business process time prediction using graph-based transformers, PGTNet performs better than the best deep learning models.
Optimizing SAGE Net: Achieving High Performance with Shorter Input Sequences for Online Inference
5 Nov 2025
This article introduces Decision Boundary-Aware Distillation, a novel method for Instance-Incremental Learning that preserves and extends the decision boundary