Recurrent Kernel Networks
This work addresses protein classification for bioinformatics, offering a novel hybrid method that combines kernel techniques with neural networks, but it is incremental as it builds on existing links between RNNs and substring kernels.
The paper tackles the problem of representing biological sequences by generalizing convolutional kernel networks to model gaps, resulting in a new type of recurrent neural network that can be trained end-to-end or without supervision. It shows that this approach outperforms existing methods for protein classification tasks, with experimental results indicating improved performance.
Substring kernels are classical tools for representing biological sequences or text. However, when large amounts of annotated data are available, models that allow end-to-end training such as neural networks are often preferred. Links between recurrent neural networks (RNNs) and substring kernels have recently been drawn, by formally showing that RNNs with specific activation functions were points in a reproducing kernel Hilbert space (RKHS). In this paper, we revisit this link by generalizing convolutional kernel networks---originally related to a relaxation of the mismatch kernel---to model gaps in sequences. It results in a new type of recurrent neural network which can be trained end-to-end with backpropagation, or without supervision by using kernel approximation techniques. We experimentally show that our approach is well suited to biological sequences, where it outperforms existing methods for protein classification tasks.