A. Globerson, G. Chechick, N. Tishby, O. Steinberg, and E. Vaadia. Distributional clustering of movements based on neural responses. unpublished manuscript, 2001.  Home/Search   Document Details and Download   Summary   Related Articles

.... approach, termed The Information Bottleneck , was developed previously by Tishby et al. in [30, 48] This approach has been used successfully in applications of text clustering [2, 30, 40, 42, 43] and astronomical observations [41] The preliminary results from its application to neural systems [14, 36] are discussed in the context of Section 4.1. The two approaches are related, due to the common term in the cost functions, I(X; Z) Z YN in our notation) and the use of probabilistic ( soft ) clustering techniques. In fact, as noted in [28] both can be seen as extensions of Grenander s ....

....only an estimate of the mutual information. We do, however, obtain a lower bound of the mutual information, albeit with higher precision. There are several drawbacks to attempting direct estimates of the joint probability, as used here in our analysis and also in applications of the direct method [14, 26, 31, 36, 45]. In principle, estimating the joint probability with a histogram is feasible only for relatively small stimulus and response spaces. For this reason a single dimensional stimulus space (stimulus angle here, arm direction in [14] and a small response space (number of spikes in a small temporal ....

[Article contains additional citation context not shown here]

A. Globerson, G. Chechick, N. Tishby, O. Steinberg, and E. Vaadia. Distributional clustering of movements based on neural responses. unpublished manuscript, 2001.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC