Unverified70% confidenceFactExact time
Elastic Weight Consolidation (EWC) protects parameters important to old tasks via the Fisher information matrix.
2
Sources
70%
Confidence
Long-term
Relevance
7/7/2026
First Seen
Sources
反向微调实验：AI如何被训练相信东京不是日本首都
hackernewshackernews7/5/2026
Related Claims
Unverified弹性权重整合（EWC）、渐进式神经网络和参数高效微调（如LoRA）被提出用于缓解灾难性遗忘71% similarUnverifiedThe mathematical form of EWC is to add a penalty term on top of the new task loss, taking the form L_new + λΣF_i(θ_i - θ_i*)^2, where θ* is a snapshot of the old task parameters, λ controls the degree of protection, and F_i comes from the Fisher information matrix.69% similarVerifiedThe three main strategies for mitigating catastrophic forgetting are regularization-based methods (such as EWC), replay-based methods (such as Experience Replay), and architecture-based methods (such as Progressive Neural Networks).59% similarUnverifiedStrategies for addressing the subtransaction performance pitfall include using SAVEPOINT judiciously, reviewing ORM framework nested transaction configurations, monitoring subtransaction overflow, and shortening the lifetime of long transactions.59% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/179093API
curl https://kongchang.com/api/v1/knowledge/claims/179093MCP
get_claim(id=179093)