How to keep the HG weights non-negative: the truncated Perceptron reweighing rule
The literature on error-driven learning in Harmonic Grammar (HG) has adopted the Perceptron reweighing rule. Yet, this rule is not suited to HG, as it fails at ensuring non-negative weights. A variant is thus considered which truncates the updates at zero, keeping the weights non-negative. Converge...
| Main Author: | |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Polish Academy of Sciences
2015-12-01
|
| Series: | Journal of Language Modelling |
| Subjects: | |
| Online Access: | https://jlm.ipipan.waw.pl/index.php/JLM/article/view/115 |
| Summary: | The literature on error-driven learning in Harmonic Grammar (HG) has adopted the Perceptron reweighing rule. Yet, this rule is not suited to HG, as it fails at ensuring non-negative weights. A variant is thus considered which truncates the updates at zero, keeping the weights non-negative. Convergence guarantees and error bounds for the original Perceptron are shown to extend to its truncated variant.
|
|---|---|
| ISSN: | 2299-856X 2299-8470 |