On-line learning and stochastic approximations
On-line learning in neural networks
Profile-Based String Kernels for Remote Homology Detection and Motif Extraction
CSB '04 Proceedings of the 2004 IEEE Computational Systems Bioinformatics Conference
Protein homology detection using string alignment kernels
Bioinformatics
A structural alignment kernel for protein structures
Bioinformatics
A hidden Markov model variant for sequence classification
IJCAI'11 Proceedings of the Twenty-Second international joint conference on Artificial Intelligence - Volume Volume Two
Input space versus feature space in kernel-based methods
IEEE Transactions on Neural Networks
| Hi-index | 0.00 |
Advances in sequencing have greatly outpaced experimental methods for determining a protein's structure and function. As a result, biologists increasingly rely on computational techniques to infer these properties of proteins from sequence information alone. We present a sequence classification framework that differs from the common SVM/kernel-based approach. We introduce a type of artificial neural network which we term the Subsequence Network (SN) that incorporates structural models over sequences in its lowest layer. These structural models, which we call Sequence Scoring Models (SSM), are similar to Hidden Markov Models and act as a mechanism to extract relevant features from sequences. In contrast to SVM/kernel methods, which only allow learning of linear discrimination weights, our feed-forward structure allows linear weights to be learned in conjunction with sequence-level features using standard optimization techniques.