This roadmap documents action items such as features or bugs to be developed/fixed.
Updated: 9 Feb 2023, 01:46 GMT
| Status | Item |
|---|---|
| ✔ | Create README and ROADMAP |
| ✔ | Implement Fisher-based unlearning such as selective synaptic dampening (Fisher forgetting not done) |
| ✔ | Implement incompetent-teacher unlearning |
| ✔ | Implement zero-glance unlearning via error-maximizing noise (UNSIR) |
| 🛑 | Implement zero-shot unlearning via gated knowledge transfer |
| ❌ | Implement Just-in-Time Unlearning via Lipschitz regularization |
| ❌ | Implement verification tests such as membership inference attacks or model inversion attacks |
| 🛑 | Implement Amnesiac Unlearning and Unlearning from AmnesiacML |
| ❌ | Implement evaluation metrics such as unlearn and relearn time, zero retrain forgetting (ZRF), anamnesis index (AIN), JS divergence, etc. |