Abstract: Some implementations are directed to generating a personal database entry for a user based on free-form natural language input formulated by the user via one or more user interface input devices of a computing device of the user. The generated personal database entry may include one or more terms of the natural language input and descriptive metadata determined based on one or more terms of the natural language input and/or based on contextual features associated with receiving the natural language input. Some implementations are directed to generating, based on one or more personal database entries of a user, output that is responsive to further free-form natural language input of the user. For example, one or more entries that are responsive to further natural language input of the user can be identified based on matching content of those entries to one or more search parameters determined based on the further input.

Abstract: Embodiments described herein provide natural language processing (NLP) systems and methods that utilize energy-based models (EBMs) to compute an exponentially-weighted energy-like term in the loss function to train an NLP classifier. Specifically, noise contrastive estimation (NCE) procedures are applied together with the EBM-based loss objectives for training the NLPs.

Abstract: Embodiments described herein provide a flexible controllable summarization system that allows users to control the generation of summaries without manually editing or writing the summary, e.g., without the user actually adding or deleting certain information under various granularity. Specifically, the summarization system performs controllable summarization through keywords manipulation. A neural network model is learned to generate summaries conditioned on both the keywords and source document so that at test time a user can interact with the neural network model through a keyword interface, potentially enabling multi-factor control.

Abstract: A method for training a probabilistic encoder-decoder having a latent space, the method including: extracting different types of medical data for a group of individuals; creating a data matrix X including the extracted medical data, wherein each row of the data matrix X includes data for one of the group of individuals; creating condition matrix C including features to define a clinical condition, wherein each row of the condition matrix C includes the condition data for one of the group of individuals; and training the encoder and the decoder to learn the latent space by minimizing the reconstruction loss and using a regularization effect to force clinically similar inputs to be close together in the latent space.

Abstract: A solid state electrolyte is provided, which includes a ligand composed of a ceramic powder and a nitrogen containing aromatic copolymer, the ceramic powder is the core and the receptor, the nitrogen containing aromatic copolymer is comprised by a first polymer and a second polymer, the first polymer is aromatic polyamide, the second polymer is selected from the group consisting of P2VP, P4VP, PVA, PEO and PAN. The solid state electrolyte can form good contact interfaces at the anode and cathode electrodes. A lithium-ion battery including the solid state electrolyte is also provided.

Abstract: Various embodiments are directed to video-based deep learning evaluation of cardiac ultrasound that accurately identify cardiomyopathy and predict ejection fraction, the most common metric of cardiac function. Embodiments include systems and methods for analyzing images obtained from an echocardiogram. Certain embodiments include receiving video from a cardiac ultrasound of a patient illustrating at least one view the patient's heart, segmenting a left ventricle in the video, and estimating ejection fraction of the heart. Certain embodiments include at least one machine learning algorithm.

Abstract: A method for monitoring health of a person by a health monitoring system includes determining if a person wearing a wearable health monitoring device is sleeping or not based on movement data, measuring sleep movements of the person to determine whether the person is in a deep sleep mode when the person is determined to be sleeping, measuring bio-vital signals of the person and at least one of a posture of the person or environmental parameters when the person is determined to be in the deep sleep mode, and automatically producing an alarm signal if a predetermined criterion is met based on the bio-vital signals and at least one of a posture of the person or the environmental parameters.

Abstract: A method for monitoring health of a person by a health monitoring system includes: measuring movement data of a person using a wearable health monitoring device, determined whether the person is in a deep sleep mode based on the movement data, automatically measuring a respiratory rate of the person that is in a deep sleep mode, determining that the person is in one of a plurality of respiration zones, measuring at least one of a body temperature, an ambient temperature, a sleeping position, a smoke or carbon monoxide level, a video, and images of the person, and automatically producing an alarm signal if a predetermined criterion is met based on the respiration zone and at least one of the body temperature, the ambient temperature, the smoke level, the carbon monoxide level, a sleeping position, or the images of the person.

Abstract: Various embodiments are directed to video-based deep learning evaluation of cardiac ultrasound that accurately identify cardiomyopathy and predict ejection fraction, the most common metric of cardiac function. Embodiments include systems and methods for analyzing images obtained from an echocardiogram. Certain embodiments include receiving video from a cardiac ultrasound of a patient illustrating at least one view the patient's heart, segmenting a left ventricle in the video, and estimating ejection fraction of the heart. Certain embodiments include at least one machine learning algorithm.

Abstract: A computer-implemented method for recognizing a user's activity pattern includes pre-storing activity data in a computer system, automatically determining locations of one or more sensors on a user's body, obtaining time series of measured activity parameters by the one or more sensors, automatically segmenting the time series of measured activity parameters into two or more activity periods, determining a spatial range of the movement in an activity period, and recognizing an activity in the activity period based at least in part on the measured activity parameters and the pre-stored activity data.

Abstract: A computer-implemented method for recognizing a user's activity pattern includes pre-storing activity data in a computer system, automatically determining locations of one or more sensors on a user's body, obtaining time series of measured activity parameters by the one or more sensors, automatically segmenting the time series of measured activity parameters into two or more activity periods, determining a spatial range of the movement in an activity period, and recognizing an activity in the activity period based at least in part on the measured activity parameters and the pre-stored activity data.