In classic pattern recognition problems, classes are mutually exclusive by deÿnition. Classiÿcation errors occur when the classes overlap in the feature space. We examine a di erent situation, occurring when the classes are, by deÿnition, not mutually exclusive. Such problems arise in semantic scene and document classiÿcation and in medical diagnosis. We present a framework to handle such problems and apply it to the problem of semantic scene classiÿcation, where a natural scene may contain multiple objects such that the scene can be described by multiple class labels (e.g., a ÿeld scene with a mountain in the background). Such a problem poses challenges to the classic pattern recognition paradigm and demands a di erent treatment. We discuss approaches for training and testing in this scenario and introduce new metrics for evaluating individual examples, class recall and precision, and overall accuracy. Experiments show that our methods are suitable for scene classiÿcation; furthermore, our work appears to generalize to other classiÿcation problems of the same nature.

Abstract-We introduce a formalism for optimal sensor parameter selection for iterative state estimation in static systems. Our optimality criterion is the reduction of uncertainty in the state estimation process, rather than an estimator-specific metric (e.g., minimum mean squared estimate error). The claim is that state estimation becomes more reliable if the uncertainty and ambiguity in the estimation process can be reduced. We use Shannon's information theory to select information-gathering actions that maximize mutual information, thus optimizing the information that the data conveys about the true state of the system. The technique explicitly takes into account the a priori probabilities governing the computation of the mutual information. Thus, a sequential decision process can be formed by treating the a priori probability at a certain time step in the decision process as the a posteriori probability of the previous time step. We demonstrate the benefits of our approach in an object recognition application using an active camera for sequential gaze control and viewpoint selection. We describe experiments with discrete and continuous density representations that suggest the effectiveness of the approach.

We present a compact and flexible endoscope (3-mm outer diameter, 4-cm rigid length) that utilizes a miniaturized resonant/nonresonant fiber raster scanner and a multielement gradient-index lens assembly for two-photon excited intrinsic fluorescence and second-harmonic generation imaging of biological tissues. The miniaturized raster scanner is fabricated by mounting a commercial double-clad optical fiber (DCF) onto two piezo bimorphs that are aligned such that their bending axes are perpendicular to each other. Fast lateral scanning of the laser illumination at 4.1 frames∕s (512 lines per frame) is achieved by simultaneously driving the DCF cantilever at its resonant frequency in one dimension and nonresonantly in the orthogonal axis. The implementation of a DCF into the scanner enables simultaneous delivery of the femtosecond pulsed 800-nm excitation source and epi-collection of the signal. Our device is able to achieve a field-of-view (FOV xy ) of 110 μm by 110 μm with a highly uniform pixel dwell time. The lateral and axial resolutions for two-photon imaging are 0.8 and 10 μm, respectively. The endoscope's imaging capabilities were demonstrated by imaging ex vivo mouse tissue through the collection of intrinsic fluorescence and second-harmonic signal without the need for staining. The results presented here indicate that our device can be applied in the future to perform minimally invasive in vivo optical biopsies for medical diagnostics.nonlinear optical endoscopy | real-time optical diagnosis | scanning fiber endoscopy | microendoscopy | endogenous fluorescence M ultiphoton imaging techniques such as two-photon fluorescence (TPF) and second-harmonic generation (SHG) microscopy hold great promise for the future of medical diagnosis because of their potential to replace surgical biopsies with minimally invasive optical diagnosis of tissue health (1-7). In a clinical setting, these diagnostic techniques will be capable of acquiring real-time, high-resolution, in vivo images without the need for contrast agents. However, a challenge in translating these beneficial imaging technologies into the clinic lies in successfully miniaturizing bulky tabletop microscope components into a compact probe without degrading the overall imaging performance of the system. These microendoscopes would not only have the potential to be used as diagnostic tools capable of early cancer detection, but could also be used for such applications as photodynamic therapy and microsurgery (8,9).A number of groups have demonstrated miniaturized instruments capable of confocal, optical coherence tomography (OCT), TPF, and SHG imaging (10-25). The primary constituents of these devices are typically a miniaturized scanning mechanism and lens assembly that is encapsulated in a protective housing with dimensions suitable for minimally invasive procedures (i.e., a probe outer diameter on the order of a few millimeters with a rigid length of several centimeters). Within these microendoscopes, various distal miniaturized scanners have been demonstr...

Context: When assessing exercise hyperthermia outdoors, the validity of certain commonly used body temperature measuring devices has been questioned. A controlled laboratory environment is generally less influenced by environmental factors (eg, ambient temperature, solar radiation, wind) than an outdoor setting. The validity of these temperature measuring devices in a controlled environment may be more acceptable.Objective: To assess the validity and reliability of commonly used temperature devices compared with rectal temperature in individuals exercising in a controlled, high environmental temperature indoor setting and then resting in a cool environment.Design: Time series study. Setting: Laboratory environmental chamber (temperature 5 36.4 6 1.26C [97.5 6 2.166F], relative humidity 5 52%) and cool laboratory (temperature 5 approximately 23.36C [74.06F], relative humidity 5 40%).Patients or Other Participants: Fifteen males and 10 females.Intervention(s): Rectal, gastrointestinal, forehead, oral, aural, temporal, and axillary temperatures were measured with commonly used temperature devices. Temperature was measured before and 20 minutes after entering the environmental chamber, every 30 minutes during a 90-minute treadmill walk in the heat, and every 20 minutes during a 60-minute rest in mild conditions. Device validity and reliability were assessed with various statistical measures to compare the measurements using each device with rectal temperature. A device was considered invalid if the mean bias (average difference between rectal and device temperatures) was more than 60.276C (60.506F).Main Outcome Measure(s): Measured temperature from each device (mean and across time).Results: Conclusions: Even during laboratory exercise in a controlled environment, devices used to measure forehead, temporal, oral, aural, and axillary body sites did not provide valid estimates of rectal temperature. Only intestinal temperature measurement met the criterion. Therefore, we recommend that rectal or intestinal temperature be used to assess hyperthermia in individuals exercising indoors in the heat.Key Words: core body temperature, hyperthermia, tympanic membrane Key Points N As indicated by mean bias values, invalid estimates of rectal temperature (the ''gold standard'') were provided by forehead sticker, oral temperature, temporal temperature, aural temperature, and axillary temperature.N Intestinal temperature was the only measurement considered valid. N Successive intestinal, forehead sticker, temporal, and aural measurements demonstrated acceptable reliability.

Marine phytoplankton account for approximately half of global primary productivity , making their fate an important driver of the marine carbon cycle. Viruses are thought to recycle more than one-quarter of oceanic photosynthetically fixed organic carbon , which can stimulate nutrient regeneration, primary production and upper ocean respiration via lytic infection and the 'virus shunt'. Ultimately, this limits the trophic transfer of carbon and energy to both higher food webs and the deep ocean . Using imagery taken by the Moderate Resolution Imaging Spectroradiometer (MODIS) onboard the Aqua satellite, along with a suite of diagnostic lipid- and gene-based molecular biomarkers, in situ optical sensors and sediment traps, we show that Coccolithovirus infections of mesoscale (~100 km) Emiliania huxleyi blooms in the North Atlantic are coupled with particle aggregation, high zooplankton grazing and greater downward vertical fluxes of both particulate organic and particulate inorganic carbon from the upper mixed layer. Our analyses captured blooms in different phases of infection (early, late and post) and revealed the highest export flux in 'early-infected blooms' with sinking particles being disproportionately enriched with infected cells and subsequently remineralized at depth in the mesopelagic. Our findings reveal viral infection as a previously unrecognized ecosystem process enhancing biological pump efficiency.