Efficient non-linear feature adaptation using Maxout networks
| Autor: | Xiaodong Cui, Vaibhava Goel, Steven J. Rennie |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
Partition function (statistical mechanics)
Feature adaptation Partition function (quantum field theory) business.industry Computer science Pattern recognition Machine learning computer.software_genre Term (time) Task (project management) 030507 speech-language pathology & audiology 03 medical and health sciences symbols.namesake Nonlinear system Jacobian matrix and determinant symbols Artificial intelligence 0305 other medical science business computer |
| Zdroj: | ICASSP |
| DOI: | 10.1109/icassp.2016.7472691 |
| Popis: | In this paper we present a simple and effective method for doing non-linear feature adaptation using Maxout networks. The technique overcomes the need to sample the partition function during training, and overcomes the need to compute the Jacobian term and its gradient for each training case. Results on the Switchboard 1 task demonstrate that the approach can improve a state-of-the-art hybrid ASR system that utilizes i-vectors. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|