Please enable JavaScript.

Coggle requires JavaScript to display documents.

RL, Multi-task Learning, Meta-learning [C], Model-based [C], Information…

- - - - often need much larger networks
    - - tasks may learn at different rates
      - caused by cross-task interference
        
        T Yu, S Kumar, PCGrad, Gradient Surgery for Multi-Task Learning, NIPS 20
- - - - External Memory
        
        Santoro et al. MANN
        
        Munkhdalai, Yu. ICML17.Meta Networks
      - Mishra et al. SNAIL, 17
- - - - Automatically learn inner vector learning rate, tune outer learning rate
        
        Li et al. Meta-SGD
        
        Behl et al. AlphaMAML
      - Optimize only a subset of the parameters in the inner loop
        
        Zhou et al. DEML
      - Decouple inner learning rate, BN statistics per-step
        
        Antoniou et al. MAML++
      - introduce context variables for increased expressive power
        
        Finn et al. bias transformation
        
        Zintgraf et al. CAVIA
    - - [CRUDELY] approximate \(\frac{d\phi_i}{d\theta}\) as identity[C]
        
        Finn et al. first-order MAML 17
        
        Nichol et al. Reptile 18
      - Only optimize the last layer of weights [C]
        
        Bertinetto et al. R2-D2 19, ridge regression, logistic regression
        
        Lee et al. MetaOptNet 19
      - Derive meta-gradient using the implicit function theorem [C]
        
        Rajeswaran, Finn, Implicit MAML 19
    - - Kim et al. Auto-Meta, progressive NAS+MAML [C]
  - - - MAML+PG
    - - Nagabandi, Learning to Adapt in Dynamic Environemtnts through Meta-RL, ICLR 19
- - - - Vinyals et al. Matching Networks, NIPS 16
      - Snell et al. Prototypical Networks, NIPS 17
  - - - Sung et al. Relation Net, learn non-linear relation module on embeddings
      - Allen et al. IMP, ICML 19, Learn infinite mixture of prototypes
      - Garcia, GNN, perform message passing on embeddings
      - Viraj et al. NIPS 18 workshop, Prototypical Clustering Networks for Dermatological Image Classification [C]
      - Triantafillou et al. Proto-MAML 19, initialize last layer as ProtoNet during meta-training
- - - - Jaderberg 17
      - Shelhamer 17
  - - - Finn, CoRL 17
      - Villegas, NIPS 19
      - Finn, Deep Visual Foresight for Planning Robot Motion, ICRA1 7
- - - - Nagabandi, Deep Dynamics Models for Learning Dexterous Manipulation, CoRL 19
    - - Xie, Few-shot Goal Inference for Visuomotor Learning and Planning, CoRL 18
- - - - common practices [C]
- - - - Zhang, MetaCURE 2020
  - - - Liu, Explore then Execute: Adopting without Rewards via Factorized Meta-RL, 2020