We present a system to recognize phrases based on perceptrons, and a global online learning algorithm to train them together. The recognition strategy applies learning in two laye...
Conventional methods for multimodal data retrieval use text-tag based or cross-modal approaches such as tag-image co-occurrence and canonical correlation analysis. Since there are ...
JungWoo Ha, Byoung-Hee Kim, Bado Lee, Byoung-Tak Z...
A popular framework for the interpretation of image sequences is based on the layered model; see e.g. Wang and Adelson [8], Irani et al. [2]. Jojic and Frey [3] provide a generativ...
Abstract. Taking neuromodulation as a mechanism underlying emotions, this paper investigates how such a mechanism can bias an artificial neural network towards exploration of new ...
—Internet application performance over wireless links is disappointing, due to wireless impairments and their adverse interactions with higher protocol layers. In order to effect...