An Evolutionary Approach to Dynamic Introduction of Tasks in Large-scale Multitask Learning Systems
This addresses the problem of scaling multitask learning for AI researchers and practitioners by enabling dynamic task addition with bounded computational costs, though it appears incremental as it builds on existing multitask and continual learning concepts.
The paper tackles the challenge of building large-scale multitask learning systems that can dynamically add new tasks without catastrophic forgetting, proposing an evolutionary method that achieves competitive results on 69 public image classification tasks, including a 15% relative error reduction on CIFAR-10 compared to the best publicly trained model.
Multitask learning assumes that models capable of learning from multiple tasks can achieve better quality and efficiency via knowledge transfer, a key feature of human learning. Though, state of the art ML models rely on high customization for each task and leverage size and data scale rather than scaling the number of tasks. Also, continual learning, that adds the temporal aspect to multitask, is often focused to the study of common pitfalls such as catastrophic forgetting instead of being studied at a large scale as a critical component to build the next generation artificial intelligence.We propose an evolutionary method capable of generating large scale multitask models that support the dynamic addition of new tasks. The generated multitask models are sparsely activated and integrates a task-based routing that guarantees bounded compute cost and fewer added parameters per task as the model expands.The proposed method relies on a knowledge compartmentalization technique to achieve immunity against catastrophic forgetting and other common pitfalls such as gradient interference and negative transfer. We demonstrate empirically that the proposed method can jointly solve and achieve competitive results on 69public image classification tasks, for example improving the state of the art on a competitive benchmark such as cifar10 by achieving a 15% relative error reduction compared to the best model trained on public data.