Research map / Continual learning / Online and replay-based
Forget Less by Learning from Parents Through Hierarchical Relationships
Arjun Ramesh Kaushik, Naresh Kumar Devulapally, Vishnu Suresh Lokhande, Nalini K. Ratha, Venu Govindaraju
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| apply hyperbolic entailment loss on image attention maps instead of LoRA weights (ablation) | FLLP · CIFC | avg Text Alignment: 74.5 -> 76.1 | Improved |
| apply hyperbolic entailment loss on image attention maps instead of LoRA weights (ablation) | FLLP · CIFC | avg Image Alignment: 79.6 -> 80.0 | No clear effect |
| add cosine-similarity guidance on task loss to CIDM (ablation) | CIDM · CIFC | avg Image Alignment: 78.0 -> 78.6 | Improved |
| add cosine-similarity guidance on task loss to CIDM (ablation) | CIDM · CIFC | avg Text Alignment: 74.8 -> 75.6 | Improved |
| replace CIDM with FLLP (hyperbolic parent entailment) | CIDM · CIFC | avg Image Alignment: 78.0 -> 80.0 | Improved |
| replace CIDM with FLLP (hyperbolic parent entailment) | CIDM · CelebA | avg Image Alignment: 73.3 -> 77.7 | Improved |
| replace CIDM with FLLP (hyperbolic parent entailment) | CIDM · ImageNet | avg Image Alignment: 81.2 -> 82.3 | Improved |
| replace CIDM with FLLP on 35 concepts | CIDM · CustomConcept101 (35 concepts) | avg Image Alignment: 71.0 -> 73.1 | Improved |
| replace CIDM with FLLP on 35 concepts | CIDM · CustomConcept101 (35 concepts) | avg Text Alignment: 75.6 -> 76.6 | Improved |
| replace CIDM with FLLP | CIDM · 1D Gaussian synthetic | forgetting rate: 13.2 -> 11.4 | Improved |
| CIDM versus vanilla continual learning | baseline continual learning · 1D Gaussian synthetic | forgetting rate: 18.6 -> 13.2 | Improved |
| replace CIDM with FLLP | CIDM · parameter drift on CustomConcept101 | avg LoRA drift: 0.018 -> 0.014 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- CLoRA — C-LoRA: Continual Low-Rank Adaptation for Pre-trained Visual Models (NeurIPS 2026) · better on 2 of 2 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Emergence of Hidden Capabilities: Exploring Learning Dynamics in Concept Space (NeurIPS 2024 · similarity 0.84)
- Learn more, but bother less: parameter efficient continual learning (NeurIPS 2024 · similarity 0.84)
- Learning Discrete Concepts in Latent Hierarchical Models (NeurIPS 2024 · similarity 0.84)
- On the Memorization of Consistency Distillation for Diffusion Models (NeurIPS 2026 · similarity 0.84)
- Fantastic Targets for Concept Erasure in Diffusion Models and Where To Find Them (ICLR 2025 · similarity 0.84)
- CASL: Concept-Aligned Sparse Latents for Interpreting Diffusion Models (NeurIPS 2026 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
