Learning to Learn Neural Networks
This work addresses the challenge of automating learning algorithms for neural networks, though it is incremental as it builds on existing meta-learning and LSTM frameworks.
The paper tackles the problem of meta-learning by using an LSTM to learn online parameter updates for another neural network, demonstrating that the learned algorithm can update parameters across layers and generalize well on similar non-linearly separable datasets.
Meta-learning consists in learning learning algorithms. We use a Long Short Term Memory (LSTM) based network to learn to compute on-line updates of the parameters of another neural network. These parameters are stored in the cell state of the LSTM. Our framework allows to compare learned algorithms to hand-made algorithms within the traditional train and test methodology. In an experiment, we learn a learning algorithm for a one-hidden layer Multi-Layer Perceptron (MLP) on non-linearly separable datasets. The learned algorithm is able to update parameters of both layers and generalise well on similar datasets.