Qualcomm Incorporated

Microsoft Technology Licensing, LLC

Bhuwan Dhingra

Shankar Sadasivam

Li Deng

Jianfeng Gao

Edward Harrison Teague

Ashwin Deepak Swaminathan

Leonard Henry Grokop

Carlos Manuel Puig

Vitor R Carvalho

Vitor Rocha De Carvalho

Tadeusz Jarosinski

Abhijeet Bisain

Lihong Li

Rajeev Jain

Xiujun Li

Ryan Michael Carey

Muralidhar Reddy Akula

Jinwon Lee

Yun-Nung Vivian Chen

Siddika Parlak Polatkan

Faisal Ahmed

Disclosed is an apparatus and method for classifying a motion state of a mobile device comprising: determining a first motion state associated with a highest probability value and with a first confidence level greater than a first threshold; entering the first motion state; while the first motion state is active, determining a second motion state associated with a highest probability value and with a second confidence level greater than the first threshold, the second motion state being different from the first motion state; determining whether the second motion state is to be entered; and in response to determining that the second motion state is to be entered, entering the second motion state.

Smartphone motion classifier

Described herein are systems, methods, and techniques by which a processing unit can build an end-to-end dialogue agent model for end-to-end learning of dialogue agents for information access and apply the end-to-end dialogue agent model with soft attention over knowledge base entries to make the dialogue system differentiable. In various examples the processing unit can apply the end-to-end dialogue agent model to a source of input, fill slots for output from the knowledge base entries, induce a posterior distribution over the entities in a knowledge base or induce a posterior distribution of a target of the requesting user over entities from a knowledge base, develop an end-to-end differentiable model of a dialogue agent, use supervised and/or imitation learning to initialize network parameters, calculate a modified version of an episodic algorithm. e.g., the REINFORCE algorithm, for training an end-to-end differentiable model based on user feedback.

End-to-end learning of dialogue agents for information access

Systems and methods herein enable a mobile device to detect that a user is traveling in association with a vehicle based at least on motion data. In some embodiments, accelerometer data is used. Motion data is leveraged in combination with various observations regarding vehicular movement to determine whether or not a mobile device is located in or on the vehicle. For instance, before entering the state of vehicular movement, it can be determined that the user is first in a walking state (e.g., walking to the car, bus, etc., and entering it). Likewise, after exiting the state of vehicular movement, the user re-enters the walking state (e.g., after stepping out of the car, bus, etc., the user again begins walking). Further, it can be determined that when the user is in the walking state, the accelerometer signals appear different to any accelerometer signals seen in the vehicular movement state.

Techniques for determination of a motion state of a mobile device

A context aware system, for use in a mobile device, includes a context change detector (CCD) coupled to a context classifier (CCL). The CCD is configured to receive sensor data and to detect a change in a current context state of the mobile device based on the received sensor data. The CCL is configured to transition from a low power consumption mode to a normal power consumption mode in response to the CCD detecting the change in the current context state. The CCL is further configured to determine a next context state of the mobile device while in the normal power consumption mode.

Context aware system with multiple power consumption modes

Growing community of inventors

New Delhi, India

Bhuwan Dhingra