Human language acquisition methods in a machine learning task

The goal of this study is to develop a psychocomputational model of human phoneme acquisition that includes the knowledge of linguistic universals [1, 2, 3] to “teach” Artificial Neural Nets incrementally. Long Short-Term Memory (LSTM) artificial neural networks are capable to outperform previous recurrent networks on many tasks ranging from grammar recognition to speech [4] and robot control [5]. Together with our psychocomputationalmodel they are supposed to recognize phonetic features in a way similar to humans learning to understand their first language.

[1]  J. Dore,et al.  Transitional phenomena in early language acquisition , 1976, Journal of Child Language.

[2]  E. Clark.  Learning how to mean: Explorations in the development of language. , 1978 .

[3]  R. Jakobson.  Child Language, Aphasia and Phonological Universals , 1980 .

[4]  Noam Chomsky,et al.  Lectures on Government and Binding , 1981 .

[5]  Jürgen Schmidhuber,et al.  Long Short-Term Memory , 1997, Neural Computation.

[6]  George Zavaliagkos,et al.  Pronunciation modeling for large vocabulary conversational speech recognition , 1998, ICSLP.

[7]  Peter Regel-Brietzmann,et al.  German regional variants - a problem for automatic speech recognition? , 1998, ICSLP.

[8]  Jürgen Schmidhuber,et al.  Learning to forget: continual prediction with LSTM , 1999 .

[9]  F. Schiel,et al.  INDEPENDENT AUTOMATIC SEGMENTATION OF SPEECH BY PRONUNCIATION MODELING , 1999 .

[10]  Tsuyoshi Ito,et al.  Generation of pronunciation rule sets for automatic segmentation of American English and Japanese , 2000, INTERSPEECH.

[11]  Juergen Schmidhuber,et al.  Long Short-Term Memory Learns Context Free and Context Sensitive Languages , 2000 .

[12]  Nicole Beringer,et al.  Regional Pronunciation Variants for Automatic Segmentation , 2000, LREC.

[13]  Bram Bakker,et al.  Reinforcement Learning with Long Short-Term Memory , 2001, NIPS.

[14]  Sepp Hochreiter,et al.  Meta-learning with backpropagation , 2001, IJCNN'01. International Joint Conference on Neural Networks. Proceedings (Cat. No.01CH37222).

[15]  Douglas Eck.  Finding downbeats with a relaxation oscillator , 2002, Psychological research.

[16]  N. Beringer.  RULE-BASED CATEGORIAL ANALYSIS OF UNPROMPTED SPEECH-A CROSS-LANGUAGE STUDY , 2003 .

[17]  Jürgen Schmidhuber,et al.  Biologically Plausible Speech Recognition with LSTM Neural Nets , 2004, BioADIT.