(Leader: Krishnaprasad)
The auditory representations described earlier provide the basic data upon which higher level decisions and classifications are made. Here we discuss the approaches we propose to investigate in incorporating auditory representations and processing as an integral part of an overall system, and specifically how to interface these ``front-end'' representations with various ``back-end'' algorithms. We focus on three proposed areas of investigation: ( A) The use of multiscale feature clustering for pattern recognition; ( B) The detection and recognition of temporal sequences; ( C) The binding of signal features using coupled oscillator networks. In all three cases, the algorithms are either inspired by auditory processes, or are particularly suited for the auditory representations at their inputs.