AutoML-Zero: Evolving Machine Learning Algorithms From Scratch
This work addresses the challenge of reducing human bias in AutoML for researchers and practitioners, though it is incremental as it builds on existing evolutionary methods.
The authors tackled the problem of automating the discovery of complete machine learning algorithms from basic mathematical operations, and demonstrated that evolutionary search can evolve algorithms like two-layer neural networks and modern techniques such as bilinear interactions and dropout-like methods, achieving results on tasks like CIFAR-10 variants.
Machine learning research has advanced in multiple aspects, including model structures and learning methods. The effort to automate such research, known as AutoML, has also made significant progress. However, this progress has largely focused on the architecture of neural networks, where it has relied on sophisticated expert-designed layers as building blocks---or similarly restrictive search spaces. Our goal is to show that AutoML can go further: it is possible today to automatically discover complete machine learning algorithms just using basic mathematical operations as building blocks. We demonstrate this by introducing a novel framework that significantly reduces human bias through a generic search space. Despite the vastness of this space, evolutionary search can still discover two-layer neural networks trained by backpropagation. These simple neural networks can then be surpassed by evolving directly on tasks of interest, e.g. CIFAR-10 variants, where modern techniques emerge in the top algorithms, such as bilinear interactions, normalized gradients, and weight averaging. Moreover, evolution adapts algorithms to different task types: e.g., dropout-like techniques appear when little data is available. We believe these preliminary successes in discovering machine learning algorithms from scratch indicate a promising new direction for the field.