Adaptation-Agnostic Meta-Training
This addresses a bottleneck in meta-learning for researchers and practitioners by enabling more flexible and powerful inner-task algorithms, though it is incremental as it builds on existing meta-learning frameworks.
The paper tackles the limitation in meta-learning where inner-task adaptation algorithms must be analytically differentiable, restricting model expressiveness, and proposes an adaptation-agnostic meta-training strategy that allows the use of stronger algorithms, achieving superior performance compared to popular baselines.
Many meta-learning algorithms can be formulated into an interleaved process, in the sense that task-specific predictors are learned during inner-task adaptation and meta-parameters are updated during meta-update. The normal meta-training strategy needs to differentiate through the inner-task adaptation procedure to optimize the meta-parameters. This leads to a constraint that the inner-task algorithms should be solved analytically. Under this constraint, only simple algorithms with analytical solutions can be applied as the inner-task algorithms, limiting the model expressiveness. To lift the limitation, we propose an adaptation-agnostic meta-training strategy. Following our proposed strategy, we can apply stronger algorithms (e.g., an ensemble of different types of algorithms) as the inner-task algorithm to achieve superior performance comparing with popular baselines. The source code is available at https://github.com/jiaxinchen666/AdaptationAgnosticMetaLearning.