Microsoft Technology Licensing, LLC

Miscrosoft Corporation

Dimitrios Lymperopoulos

Jie Jerry Liu

Nissanka Arachchige Bodhi Priyantha

Douglas Christopher Burger

Feng Zhao

Di Wang

Yuri Pekelny

Melissa Wood Dunn

Pedro Urbina Escos

Emanuel Shalev

Gerald Reuben DeJean

Fang Wang

Nissanka Bodhi Priyantha

Ashwini Kumar Varma

Deepak Ganesan

Jen-Hsien Kenny Chien

A computer implemented method includes obtaining a first deep neural network (DNN) model trained on labeled real image data for a downstream vision task, obtaining a second DNN model trained on synthetic images created with random image parameter values for the downstream vision task, obtaining a third DNN model trained on the labeled real image data and the synthetic images for the downstream vision task, performing a forward pass execution of each model to generate a loss, backpropagating the loss to modify parameter values, and iterating the forward pass execution and backpropagating with images generated by the modified parameters to jointly train the models and optimize the parameters.

Model-aware synthetic image generation

In general, the short range wireless powered ring described herein pertains to a finger-worn ring. In one embodiment the ring has a small interaction area towards the palm of the hand that employs sensors to sense user input and can interpret this input and other data. For example, the ring can interpret user input as scroll and select input actions. The ring communicates user interactions and other data wirelessly using a low-power wireless solution. The ring contains a coil and other circuitry for energy harvesting from short range wireless enabled devices such as, for example, NFC enabled phones, while users interact with their devices. A built in rechargeable battery is used to store the scavenged energy. The ring may also contain physiological and inertial sensors. The ring can provide a readily available input device of small form factor that has an easily accessible energy source for ease of use.

Short range wireless powered ring for user interaction and sensing

A location-related entity ranking technique is described that allows a user see the ranking of location-related entities near a desired location in response to a query. The technique can leverage the searches performed by a community of mobile device users at a given location and at a given time to infer the location-related entities that are popular at this specific location and time window. The technique can also filter out or re-rank these location-related entities based on the personal preferences of the person that submits the query and group preferences of a group to which the user belongs. The personal preferences can also be context specific (i.e. different restaurant preferences for lunch and dinner times).

Real-time personalized recommendation of location-related entities

A 'Communication Device with Energy Efficient Sensing' provides various techniques for enabling a variety of sensor-based applications and processes while conserving energy on mobile communications devices. More specifically, the Communication Device with Energy Efficient Sensing provides a novel sensing architecture for mobile communications devices where sampling and, in various embodiments, processing, of sensor data received from one or more sensors is offloaded to a dedicated low-power processor having the capability to interact with a primary processor of the communications device. Such sensors include, but are not limited to accelerometers, ambient light sensors, compasses, microphones, pressure sensors, touch sensors, low-power radio devices, etc. This approach enables the communications device to perform continuous sensing with a low power overhead. In various tested embodiments, it has been observed that that the sensing architecture described herein can be two to three orders of magnitude more energy efficient in comparison to conventional communications devices.

Energy efficient continuous sensing for communications devices

A smart interconnect for modular multi-component embedded devices is described. In an embodiment of a smart interconnect for modular multi-component embedded devices, a desired functionality of a stack of hardware boards is accessed. For example, a user may select a new functionality for the stack of hardware boards. The desired functionality is then transmitted to a board in the stack of hardware boards and the board is configured to implement the desired functionality of the stack of hardware boards.

Smart interconnect for modular multi-component embedded devices

Growing community of inventors

Bellevue, WA, United States of America

Dimitrios Lymperopoulos