Towards a Mobile Multimedia Age – Location-Based Services: A Case Study

Japan has been at the leading edge of location-based services (LBS) deployment using wireless assisted GPS technology. These services, which include personal security, navigation, gaming, security and fleet management, have experienced tremendous growth over the last few years. By looking at the factors contributing to this successful growth, LBS can be deployed in Europe with equal success. In addition to success factors, it is important to look at the underlying wireless network technology and its deployment impact, since Japanese deployments are based on synchronous 3G whereas Europe's likely deployments will be based on asynchronous UMTS. Recognizing the implication of both these items in accordance with SnapTrack's vision of UMTS LBS deployment affords a service provider a better opportunity to successfully introduce LBS and drive service revenue. This revised version was published online in August 2006 with corrections to the Cover Date.     

Relation of olfactory EEG to behavior: Factor analysis.

Rabbits were conditioned to lick (CR+) in response to one odor (CS+); another odor (CS–) served as a discriminative control (CR–). Electroencephalograms (EEGs) were recorded from arrays of 64 electrodes on the olfactory bulb in three stages, each with six sessions: in Stage I, odors A+ and B–; in Stage II, odors C+ and B–; and in Stage III, odors C+ and A–. Spatial EEG amplitude patterns were measured for multiple control (C), CS+, and CS– EEG bursts in each trial. Data were transformed via factor analysis and expressed by factor scores as spatial patterns specified by factor loadings. In discriminant analysis of the factor scores, we correctly classified the C and CS bursts on the average by 65–80% from all trials for each subject and session and by 75–90% for trials with correct CRs. The latter was confirmed with a stepwise linear discriminant analysis of the original 64-variable data. Factor patterns were relatively invariant within but changed between stages. The results implied that stable spatial patterns of bulbar activity emerged in respect to CSs under reinforcement and persisted until the stimulus-response contingencies were changed. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Spatial EEG correlates of nonassociative and associative olfactory learning in rabbits.

Recent studies have shown that spatially distributed olfactory bulbar activity correlates with odor-specific behavioral responding (e.g., W. J. Freeman and K. A. Grajski [see PA, Vol 75:28775]; Freeman and W. Schneider [see PA, Vol 68:5257]). The present studies established olfactory bulbar spatial EEG correlates of nonassociative and associative learning in odorant stimulation in rabbits. Behavior was quantified by measuring magnitude and probability of the sniff response. It was shown that (a) olfactory bulbar spatial EEG amplitude patterns do not simply reflect odor (peripheral) stimulation, (b) repeated presentations of a nonreinforced odor initially reveal a transient EEG pattern change but the pattern change does not recur after the subject has habituated to the odor, and (c) repeated presentations of a reinforced odor (mild cutaneous shock), with a second nonreinforced odor serving as a control, reveal that coexisting, odorspecific spatial EEG amplitude patterns emerge with the acquisition of differential behavioral responding. (PsycINFO Database Record (c) 2010 APA, all rights reserved)     

Classification of EEG Spatial Patterns with a Tree-Structured Methodology: CART

The increasing use of computers in statistics has spawned a new generation of multivariate statistical techniques. Chief among these is a tree-structured approach to classification and regression analysis. The CART, or Classification and Regression Trees, program implements a recursive partitioning procedure based on an iterative search for best binary "splits" of data. Resultant classifiers consist of binary trees whose leaves determine class labeling. Extensive use of data resampling techniques replaces biased classifier performance measures.