RIKEN Tensor Decomposition Incremental Learning
https://icml.cc/virtual/2025/poster/44196
Uses tensor decomposition to exploit low intrinsic dimensionality and pixel correlation of stored exemplar images, achieving high compression while preserving discriminative information for class-incremental learning. Compresses the replay buffer used in catastrophic-forgetting prevention. RIKEN AIP + Guangdong U. of Technology.
At a glance
- Type
- Tensor-decomposed exemplar compression
- Tier
- T3
- Created
- 2025 (ICML 2025 poster; exact arxiv submission date not confirmed)
- Latest release
- not applicable — not OSS
- License
- not applicable — not OSS
- GitHub
- not applicable — no GitHub repo
- Pricing
- not applicable — not commercial
- Funding
- not applicable — not commercial
Taxonomy
- storage
- parametric
- retrieval
- parametric-recall
- persistence
- parametric-permanent
- update
- parametric-edit
- unit
- episode
- governance
- n/a
- conflict
- n/a
When to use
Optimised for: not applicable - research paper
Anti-fit: not applicable - research paper
Pros & cons
Pros
Principled mathematical basis (tensor low-rank structure) rather than heuristic pruning; applies to any replay-based continual learning setup.
Cons
Targets image classification exemplar buffers specifically; applicability to language model continual learning is indirect and not demonstrated in this paper.
Claims & capabilities
ICML 2025 poster; compression without proportional accuracy loss vs full-exemplar replay.
Technical surface
- API surface
- not applicable — research paper
- Backend storage
- not applicable — research paper
- Deployment
- not applicable — not a deployable product
- Embedding model
- not applicable — research paper
- Multi-tenancy
- not applicable — research paper
- MCP
- not applicable — research paper, no deployed product
- A2A
- not applicable — research paper, no deployed product
- OpenTelemetry
- not applicable — research paper, no deployed product
Similar systems
Other recent method papers — theorized, no distinct product in the catalog, ranked by inbound references.
- Compressive Transformer T3
Maintains recent states in full resolution while compressing older memories with learned compression functions. DeepMind.
- MemGPT v2 / agent-tools T3
Already in catalog as the foundational MemGPT paper. Note: Letta is the productionised successor (cross-listed).
- Transformer-XL T3
Extends context through segment-level recurrence + caching of hidden states from prior segments. Foundational long-context architecture.
- Generative Agents T3
Park et al. — landmark agent-simulation paper. Reflection + memory stream + retrieval enable believable agent behavior.
- MemoryBank T3
Enhances LLMs with long-term memory. Early influential paper.
- Reflexion T3
Language agents with verbal reinforcement learning.