Retinal adaptation of visual processing time delays.

A significant proportion of the processing delays within the visual system are luminance dependent. Thus placing an attenuating filter over one eye causes a temporal delay between the eyes and thus an illusion of motion in depth for objects moving in the fronto-parallel plane, known as the Pulfrich effect. We have used this effect to study adaptation to such an interocular delay in two normal subjects wearing 75% attenuating neutral density filters over one eye. In two separate experimental periods both subjects showed about 60% adaptation over 9 days. Reciprocal effects were seen on removal of the filters. To isolate the site of adaptation we also measured the subjects' flicker fusion frequencies (FFFs) and contrast sensitivity functions (CSFs). Both subjects showed significant adaptation in their FFFs. An attempt to model the Pulfrich and FFF adaptation curves with a change in a single parameter in Kelly's [(1971) Journal of the Optical Society of America, 71, 537-546] retinal model was only partially successful. Although we have demonstrated adaptation in normal subjects to induced time delays in the visual system we postulate that this may at least partly represent retinal adaptation to the change in mean luminance.

Learning optimal adaptation strategies in unpredictable motor tasks.

Picking up an empty milk carton that we believe to be full is a familiar example of adaptive control, because the adaptation process of estimating the carton's weight must proceed simultaneously with the control process of moving the carton to a desired location. Here we show that the motor system initially generates highly variable behavior in such unpredictable tasks but eventually converges to stereotyped patterns of adaptive responses predicted by a simple optimality principle. These results suggest that adaptation can become specifically tuned to identify task-specific parameters in an optimal manner.

Bayesian adaptation and adaptively trained systems

As the use of found data increases, more systems are being built using adaptive training. Here transforms are used to represent unwanted acoustic variability, e.g. speaker and acoustic environment changes, allowing a canonical model that models only the "pure" variability of speech to be trained. Adaptive training may be described within a Bayesian framework. By using complexity control approaches to ensure robust parameter estimates, the standard point estimate adaptive training can be justified within this Bayesian framework. However during recognition there is usually no control over the amount of data available. It is therefore preferable to be able to use a full Bayesian approach to applying transforms during recognition rather than the standard point estimates. This paper discusses various approximations to Bayesian approaches including a new variational Bayes approximation. The application of these approaches to state-of-the-art adaptively trained systems using both CAT and MLLR transforms is then described and evaluated on a large vocabulary speech recognition task. © 2005 IEEE.