Training and search methods for speech recognition.

Speech recognition involves three processes: extraction of acoustic indices from the speech signal, estimation of the probability that the observed index string was caused by a hypothesized utterance segment, and determination of the recognized utterance via a search among hypothesized alternatives. This paper is not concerned with the first process. Estimation of the probability of an index string involves a model of index production by any given utterance segment (e.g., a word). Hidden Markov models (HMMs) are used for this purpose [Makhoul, J. & Schwartz, R. (1995) Proc. Natl. Acad. Sci. USA 92, 9956-9963]. Their parameters are state transition probabilities and output probability distributions associated with the transitions. The Baum algorithm that obtains the values of these parameters from speech data via their successive reestimation will be described in this paper. The recognizer wishes to find the most probable utterance that could have caused the observed acoustic index string. That probability is the product of two factors: the probability that the utterance will produce the string and the probability that the speaker will wish to produce the utterance (the language model probability). Even if the vocabulary size is moderate, it is impossible to search for the utterance exhaustively. One practical algorithm is described [Viterbi, A. J. (1967) IEEE Trans. Inf. Theory IT-13, 260-267] that, given the index string, has a high likelihood of finding the most probable utterance.

Dynamic weighting in Monte Carlo and optimization

Dynamic importance weighting is proposed as a Monte Carlo method that has the capability to sample relevant parts of the configuration space even in the presence of many steep energy minima. The method relies on an additional dynamic variable (the importance weight) to help the system overcome steep barriers. A non-Metropolis theory is developed for the construction of such weighted samplers. Algorithms based on this method are designed for simulation and global optimization tasks arising from multimodal sampling, neural network training, and the traveling salesman problem. Numerical tests on these problems confirm the effectiveness of the method.

Adrenocortical suppression blocks the memory-enhancing effects of amphetamine and epinephrine.

This study examined glucocorticoid-adrenergic interactions in modulating acquisition and memory storage for inhibitory avoidance training. Systemically (s.c.) administered amphetamine (1 mg/kg), but not epinephrine (0.1 mg/kg) or the peripherally acting amphetamine derivative 4-OH amphetamine (2 mg/kg), given to rats shortly before training facilitated acquisition performance in a continuous multiple-trial inhibitory avoidance (CMIA) task. Adrenocortical suppression with the 11beta-hydroxylase inhibitor metyrapone (50 mg/kg; s.c.), given to rats 90 min before training, did not block the effect of amphetamine and did not affect acquisition performance of otherwise untreated animals. Retention of CMIA and one-trial inhibitory avoidance was enhanced by either pre- or posttraining injections of amphetamine as well as 4-OH amphetamine and epinephrine. The finding that injections of amphetamine and epinephrine have comparable effects on memory is consistent with the view that amphetamine may modulate memory storage, at least in part, by inducing the release of epinephrine from the adrenal medulla. Metyrapone pretreatment blocked the memory-enhancing effects of amphetamine, 4-OH amphetamine, and epinephrine but did not affect retention performance of otherwise untreated animals. Posttraining injections of different doses of epinephrine (ranging from 0.0001 to 1.0 mg/kg) produced a dose-dependent memory enhancement for inhibitory avoidance training and metyrapone blocked the memory-enhancing effects of all these doses. These findings provide further evidence that the sympathoadrenal and adrenocortical systems are intimately coupled during processes of memory storage.