Cross-channel degradation is one of the significant challenges facing speaker recognition systems. We study the problem for speaker recognition using support vector machines (SVMs). We perform channel compensation in SVM modeling by removing non-speaker nuisance dimensions in the SVM expansion space via projections. Training to remove these dimensions is accomplished via an eigenvalue problem. The eigenvalue problem attempts to reduce multisession variation for the same speaker, reduce different channel effects, and increase "distance" between different speakers. We apply our methods to a subset of the Switchboard 2 corpus. Experiments show dramatic improvement in performance for the cross-channel case.

What is the neural representation of a speech code as it evolves in time? A neural model simulates data concerning segregation and integration of phonetic percepts. Hearing two phonetically related stops in a VC-CV pair (V = vowel; C = consonant) requires 150 ms more closure time than hearing two phonetically different stops in a VC,-C 2 V pair. Closure time also varies with long-term stimulus rate. The model simulates rate-dependent category boundaries that emerge from feedback: interactions between a working memory for short-term storage of phonetic items and a list categorization network for grouping sequences of items. The conscious speech code is a resonant wave. It emerges after bottom-up signals from the working memory select list chunks which read out top-down expectations that amplify and focus attention on consistent working memory items. In VCi-C 2 V pairs, resonance is reset by mismatch of Cj with the C, expectation. In VC-CV pairs, resonance prolongs a repeated C.What is the nature of the process that converts brain events into behavioral percepts? An answer to this question is needed in order to understand how the brain controls behavior and how the brain is, in turn, shaped by environmental feedback that is experienced on the behavioral level. The nature of this connection also needs to be understood in order to develop neurally plausible connectionist models. Without it, a correct linking hypothesis cannot be developed between psychological data and the brain mechanisms from which they are generated. The Resonant Dynamics of Conscious Speech PerceptsThe present article illustrates the hypothesis that conscious speech percepts are emergent properties that arise from resonant states of the brain. Such a resonance develops when bottom-up signals that are activated by environmental events interact with top-down expectations, or prototypes, that have been learned from prior experiences. The topdown expectations carry out a matching process that selects those combinations of bottom-up features that are consistent with the learned prototype while inhibiting those that are We wish to thank Diana Meyers for her invaluable assistance in the preparation of the article.Correspondence concerning this article should be addressed to Stephen Grossberg, Department of Cognitive and Neural Systems and Center for Adaptive Systems, Boston University, 677 Beacon Street, Boston, Massachusetts 02215. Electronic mail may be sent via Internet to steve@cns.bu.edu. not. In this way, an attentional focus starts to develop that concentrates processing on those feature clusters that are deemed important on the basis of past experience. The attended feature clusters, in turn, reactivate the cycle of bottom-up and top-down signal exchange. This reciprocal exchange of signals eventually equilibrates in a resonant state that binds the attended features together into a coherent brain state. Such resonant states, rather than the activations that are due to bottom-up processing alone, are proposed to be the brain events that represen...

Neurons from layer II of the medial entorhinal cortex show subthreshold membrane potential oscillations (SMPOs) which could contribute to theta-rhythm generation in the entorhinal cortex and to generation of grid cell firing patterns. However, it is unclear if single neurons have a fixed unique oscillation frequency or if their frequency varies depending on the mean membrane potential in a cell. We therefore examined the frequency of SMPOs at different membrane potentials in layer II stellate-like cells of the rat medial entorhinal cortex in vitro. Using whole cell patch recordings, we found that the fluctuations in membrane potential show a broad band of low power frequencies near resting potential that transition to more narrow band oscillation frequencies with depolarization. The transition from broadband to narrow band frequencies depends on the location of the neuron along the dorso-ventral axis in the entorhinal cortex, with dorsal neurons transitioning to higher frequency oscillations relative to ventral neurons transitioning to lower frequency oscillations. Once SMPOs showed a narrow band frequency, systematic frequency changes were not observed with further depolarization. Using a Hodgkin-Huxley style model of membrane currents we show that differences in the influence of depolarization on the frequency of SMPOs at different dorsal to ventral positions could arise from differences in the properties of the h current. The properties of frequency changes in this data are important for evaluating models of the generation of grid cell firing fields with different spacings along the dorsal to ventral axis of medial entorhinal cortex.

The resonance properties of individual neurons in entorhinal cortex may contribute to their functional properties in awake, behaving rats. Models propose that entorhinal grid cells could arise from shifts in the intrinsic frequency of neurons caused by changes in membrane potential due to depolarizing input from neurons coding velocity. To test for potential changes in intrinsic frequency we measured the resonance properties of neurons at different membrane potentials in neurons in medial and lateral entorhinal cortex. In medial entorhinal neurons the resonant frequency of individual neurons decreased in a linear manner as the membrane potential was depolarized between −70 and −55 mV. At more hyperpolarized membrane potentials, cells asymptotically approached a maximum resonance frequency. Consistent with previous studies, near resting potential the cells of the medial EC possessed a decreasing gradient of resonance frequency along the dorsal to ventral axis, and cells of the lateral EC lacked resonant properties regardless of membrane potential or position along the medial to lateral axis within lateral EC. Application of 10 µM ZD7288, the H-channel blocker, abolished all resonant properties in MEC cells, and resulted in physiological properties very similar to lateral EC cells. These results on resonant properties show a clear change in frequency response with depolarization that could contribute to generation of grid cell firing properties in the medial entorhinal cortex.

How does the brain extract invariant properties of variable-rate speech? A neural model, called PHONET, is developed to explain aspects of this process and, along the way, data about perceptual context effects. For example, in consonant vowel (CV) syllables such as /ba/ and /wa/, an increase in the duration of the vowel can cause a switch in the percept of the preceding consonant from /w I to /b/ (Miller and Liberman, 1979).The frequency extent of the initial formant transitions of fixed duration also influences the percept (Schwab, Sawusch, and Nusbaum, 1981). PHONET quantitatively simulates over 98% of the variance in these data using a single set of parameters. The model also qualitatively explains many data about other perceptual context effects. In the model, Brain analogs of model cell types are described.

No abstract

We studied the Near-Earth Asteroid (NEA) candidates posted on the Minor Planet Center's Near-Earth Object Confirmation Page (NEOCP) between years 2013 and 2016. Out of more than 17,000 NEA candidates, while the majority became either new discoveries or were associated with previously known objects, about 11% were unable to be followed-up or confirmed. We further demonstrate that of the unconfirmed candidates, 926 ± 50 are likely to be NEAs, representing 18% of discovered NEAs in that period. Only 11% (∼ 93) of the unconfirmed NEA candidates were large (having absolute magnitude H < 22). To identify the reasons why these NEAs were not recovered, we analyzed those from the most prolific asteroid surveys: Pan-STARRS, the Catalina Sky Survey, the Dark Energy Survey, and the Space Surveillance Telescope. We examined the influence of plane-of-sky positions and rates of motion, brightnesses, submission delays, and computed absolute magnitudes, as well as correlations with the phase of the moon and seasonal effects. We find that delayed submission of newly discovered NEA candidate to the NEOCP drove a large fraction of the unconfirmed NEA candidates. A high rate of motion was another significant contributing factor. We suggest that prompt submission of suspected NEA discoveries and rapid response to fast moving targets and targets with fast growing ephemeris uncertainty would allow better coordination among dedicated 2 Vereš et al.follow-up observers, decrease the number of unconfirmed NEA candidates, and increase the discovery rate of NEAs.

No abstract