Alex G. Lee (email@example.com)
Deep learning is a machine learning method that attempts to learn layered models of inputs that mimics the human brain’s reasoning process. The layers correspond to distinct levels of concepts where higher-level concepts are derived from lower-level concepts (hierarchy of complex concepts that are constructed out of simpler concepts).
US20150294422 illustrates the deep learning application for dynamic risk management in the IoT connected cars. The deep learning correlates the driver's usage patterns, driving styles, weather conditions, traffic or interaction with the vehicle based on a priori knowledge of how normalized driver data sets behave to predict potential driving risk dynamically. Real time data are provided using the vehicle-to-infrastructure (V2I) and vehicle-to-vehicle (V2V) communications. US20150166072 illustrates another deep learning application for dynamic risk management in the IoT connected cars. The deep learning correlates multiple vehicles behavior as a function of their environment (e.g., road conditions pertain to an individual vehicle) in order to predict vehicle behavior in future environments.
Evaluation of a living environment such as an interior space highly influences residents and users of the space, the space itself, and uses and purposes of the space. US20140241627 illustrates the deep learning application for evaluating the effect of the lighting on human feelings in the IoT smart home. The deep learning correlates the brightness of the overall space and human feelings toward images to classify the human feeling categories. The classification can be used to provide a better smart home lighting system. US20150286638 illustrates another deep learning application for recognizing the scene type from a set of images in the IoT smart home. The deep learning correlates the set of images to learn new features. The scene recognizing system can be used to provide the context-aware smart home services.
Distributed power generation refers to a mode of power generation in which electricity is generated by energy sources at locations distributed across the power grid. US20140340236 illustrates the deep learning application for securing the distributed power distribution networks in the IoT smart grids. The basic process in the security surrounding the power distribution networks involves the notion of discriminating between normal operations and anomalous operations. The deep learning can be used to discriminate between a signature considered normal, and one that is abnormal, or as well as sub-classes of abnormality due to malicious or non-malicious perturbations of the networks. The deep learning is trained on a training set of timing of communication input labeled as "normal." Sets of abnormal input can be labeled and used to train the deep learning to discriminate particular classes of abnormal signatures.
US20150142466 illustrates the deep learning application for providing remote medical diagnosis and therapy of the IoT connected healthcare. The deep learning predicts a risk of mortality and/or morbidity based on the statistical modeling of health outcomes from historic and real-time health data.
©2015 TechIPm, LLC All Rights Reserved http://www.techipm.com/