Gaussian Prototypical Networks for Few-Shot Learning on Omniglot
This work addresses few-shot learning for AI systems by introducing an incremental improvement that enhances robustness in noisy, real-world datasets.
The authors tackled few-shot classification on Omniglot by proposing Gaussian prototypical networks, which incorporate uncertainty estimates as Gaussian covariances to weight distances in embedding space, achieving state-of-the-art performance with improvements in 1-shot and 5-shot tasks, such as comparable results in 5-shot 5-way classification.
We propose a novel architecture for $k$-shot classification on the Omniglot dataset. Building on prototypical networks, we extend their architecture to what we call Gaussian prototypical networks. Prototypical networks learn a map between images and embedding vectors, and use their clustering for classification. In our model, a part of the encoder output is interpreted as a confidence region estimate about the embedding point, and expressed as a Gaussian covariance matrix. Our network then constructs a direction and class dependent distance metric on the embedding space, using uncertainties of individual data points as weights. We show that Gaussian prototypical networks are a preferred architecture over vanilla prototypical networks with an equivalent number of parameters. We report state-of-the-art performance in 1-shot and 5-shot classification both in 5-way and 20-way regime (for 5-shot 5-way, we are comparable to previous state-of-the-art) on the Omniglot dataset. We explore artificially down-sampling a fraction of images in the training set, which improves our performance even further. We therefore hypothesize that Gaussian prototypical networks might perform better in less homogeneous, noisier datasets, which are commonplace in real world applications.