Abstract: In some examples, a data management and storage (DMS) system comprises peer DMS nodes in a node cluster, a distributed data store comprising local and cloud storage, and an IO request scheduler comprising at least one processor configured to perform operations in a method of scheduling IO requests. Example operations comprise implementing a kernel scheduler to schedule a flow of IO requests in the DMS system, and providing an adjustment layer to adjust the kernel scheduler based on an IO request prioritization. A flow of IO requests is identified and some examples implement an IO request prioritization based on the adjustments made by the adjustment layer.

Abstract: The present disclosure describes systems and methods for implementing input-adaptive neural networks. Several approaches and systems are presented for creating flexible and computationally-efficient deep neural networks that adapt their complexity based upon input data such as video streams. The approach may use fine-grained design and an automatic approach for generating optimal network architectures based on early-exit mechanisms. A neural network modified according to such principles can be implemented in various devices including mobile vision systems such as smartphones, drones, and head-mounted augmented reality devices.

Abstract: A system may perform speech enhancement of audio data in real-time by suppressing noise components that are present in the audio data while preserving speech components. The system may include an in-ear module and a separate signal processing module that is wirelessly communicatively coupled to the in-ear module. The system may include non-negative matrix factorization (NMF) dictionaries capable of identifying frequency band components associated with speech and frequency band components associated with noise. The NMF dictionaries may be trained using voice samples and noise samples. The NMF dictionaries may be applied to noisy speech data to produce an NMF representation of the speech data which may then be applied using a dynamic mask to the noisy speech data in order to suppress the noise components of the noisy speech data and produce speech enhanced data.

Abstract: Techniques for distributed computing system node management are described herein. In some cases, internal compute nodes (i.e., compute nodes that are allocated to the distributed system) may be mutually trusted such that they may freely establish communications with one another. By contrast, external compute nodes (i.e., compute nodes that aren't allocated to the distributed computing system) may be untrusted such that their access to the distributed system may be regulated. In some cases, one or more of the compute nodes within the distributed computing system may maintain respective collections of system view information. Each respective collection of system view information may include, for example, information associated with the corresponding compute node's view of the distributed computing system based on information that is available to the corresponding compute node.

Abstract: A system may perform speech enhancement of audio data in real-time by suppressing noise components that are present in the audio data while preserving speech components. The system may include an in-ear module and a separate signal processing module that is wirelessly communicatively coupled to the in-ear module. The system may include non-negative matrix factorization (NMF) dictionaries capable of identifying frequency band components associated with speech and frequency band components associated with noise. The NMF dictionaries may be trained using voice samples and noise samples. The NMF dictionaries may be applied to noisy speech data to produce an NMF representation of the speech data which may then be applied using a dynamic mask to the noisy speech data in order to suppress the noise components of the noisy speech data and produce speech enhanced data.

Abstract: Techniques for distributed computing system node management are described herein. In some cases, internal compute nodes (i.e., compute nodes that are allocated to the distributed system) may be mutually trusted such that they may freely establish communications with one another. By contrast, external compute nodes (i.e., compute nodes that aren't allocated to the distributed computing system) may be untrusted such that their access to the distributed system may be regulated. In some cases, one or more of the compute nodes within the distributed computing system may maintain respective collections of system view information. Each respective collection of system view information may include, for example, information associated with the corresponding compute node's view of the distributed computing system based on information that is available to the corresponding compute node.

Abstract: A system and method is provided that utilizes deep learning, including convolutional neural networks, to identify subject objects in unconstrained user images such as unknown pills. An image of, e.g., a pill, may be captured and subsequently processed using deep learning models to identify the pill. The deep learning models may be optimized to have a small footprint (in terms of computational and memory resources) suitable for a resource-limited device such as a smartphone while retaining a high object recognition accuracy. Each such model may also be run on modified versions of the unconstrained image, for example on color, greyscale, and gradient images, to focus the models on different distinguishing features of the object.

Abstract: Techniques for distributed computing system node management are described herein. In some cases, internal compute nodes (i.e., compute nodes that are allocated to the distributed system) may be mutually trusted such that they may freely establish communications with one another. By contrast, external compute nodes (i.e., compute nodes that aren't allocated to the distributed computing system) may be untrusted such that their access to the distributed system may be regulated. In some cases, one or more of the compute nodes within the distributed computing system may maintain respective collections of system view information. Each respective collection of system view information may include, for example, information associated with the corresponding compute node's view of the distributed computing system based on information that is available to the corresponding compute node.

Abstract: A system may perform speech enhancement of audio data in real-time by suppressing noise components that are present in the audio data while preserving speech components. The system may include an in-ear module and a separate signal processing module that is wirelessly communicatively coupled to the in-ear module. The system may include non-negative matrix factorization (NMF) dictionaries capable of identifying frequency band components associated with speech and frequency band components associated with noise. The NMF dictionaries may be trained using voice samples and noise samples. The NMF dictionaries may be applied to noisy speech data to produce an NMF representation of the speech data which may then be applied using a dynamic mask to the noisy speech data in order to suppress the noise components of the noisy speech data and produce speech enhanced data.

Abstract: The double pulse generator of the level shifter circuit takes out the rising edge and falling edge of the pulse width modulation signal PWM_H and generates corresponding narrow pulse signals. The two narrow pulse signals respectively pass through the pulse shaper to control the two field effect transistors in the switching circuit. The pulse width of the narrow pulse signal is not enough to completely switch on the two field effect transistors, so the generated waveform is a sawtooth wave; the drains of the two field effect transistors are respectively connected to the hysteresis-adjustable Schmidt trigger to restore the narrow pulse signal to the rising edge and falling edge pulse signal of the pulse width modulation signal PWM_HS with respect to the floating side VS, and then the signal is restored to the level-shifted pulse width modulation signal PWM HS after passing through the RS latch.

Abstract: A gate charge and discharge adjustment regulating circuit for a gate control device belongs to the power electronics technology field. The switch control signal is connected to the control terminals of the four analog switches. The gate control signal is loaded on the gate of the correct field effect transistor under the action of the four analog switches to control the switching-on degree so as to achieve the purpose of adjusting the gate driving signal current, that is, regulating the gate charge and discharge currents of the gate control device to realize the change of the switching characteristics and conduction characteristics. The switch control signal is connected to the input terminal of the gate driving module to control the gate driving module to generate the gate driving signal.

Abstract: A system and method is provided that utilizes deep learning, including convolutional neural networks, to identify subject objects in unconstrained user images such as unknown pills. An image of, e.g., a pill, may be captured and subsequently processed using deep learning models to identify the pill. The deep learning models may be optimized to have a small footprint (in terms of computational and memory resources) suitable for a resource-limited device such as a smartphone while retaining a high object recognition accuracy. Each such model may also be run on modified versions of the unconstrained image, for example on color, greyscale, and gradient images, to focus the models on different distinguishing features of the object.

Abstract: An example apparatus is caused to receive a video sequence of a plurality of frames, and perform a number of operations as each of at least some of the frames is received but before all of the frames are received. The apparatus is caused to calculate a score for the frame, and compare the score for the frame to a predefined threshold. The apparatus is caused to cause output of the frame as a key frame in an instance in which the frame is received within a specified period of time and the score for the frame is above the predefined threshold. Otherwise, in an instance in which none of the scores for frames received within the specified period of time is above the predefined threshold, the apparatus is caused to cause output of one of the frames received within the specified period of time as a key frame.

Abstract: The methods provided focus on a quantitative molecular assay tools that systematically measure a set of pre-selected targets, with proper controls in a biological sample for identification of biomarkers or novel targets for a disease status. This allows for systematically maximizing the power of multivariate feature selection tools on the analysis of high-throughput screening data (such as microarray) and use of the well selected target to generate a qPCR array with tissue specific controls and qPCR controls to serve the needs of biomarker study.

Abstract: The methods provided herein use microarray data for feature selection and then use selected targets to generate industry standard qPCR arrays with new clinical sample assay data so order to build a classification model. This multi-step method overcomes the disadvantages of traditional biomarker identification.

Abstract: An approach is provided for enhancing a communication session. The approach involves receiving a request for communicating at least substantially live video data between a device and one or more other devices. The approach also involves determining capability information, resource availability information, or a combination thereof of the device, the one or more other devices, or a combination thereof. The approach further involves processing and/or facilitating a processing of the capability information, the resource availability information, or a combination thereof to cause, at least in part, an extraction of multimodal information from the at least substantially live video data. The approach additionally involves causing, at least in part, an exchange of the multimodal information in place of at least a portion of the at least substantially live video data in response to the request.

Abstract: An example apparatus is caused to receive a video sequence of a plurality of frames, and perform a number of operations as each of at least some of the frames is received but before all of the frames are received. The apparatus is caused to calculate a score for the frame, and compare the score for the frame to a predefined threshold. The apparatus is caused to cause output of the frame as a key frame in an instance in which the frame is received within a specified period of time and the score for the frame is above the predefined threshold. Otherwise, in an instance in which none of the scores for frames received within the specified period of time is above the predefined threshold, the apparatus is caused to cause output of one of the frames received within the specified period of time as a key frame.

Abstract: The present invention provides conjugates between peptides and PEG moieties through glycerol linkers.

Abstract: This invention provides gene-specific probe labeling methods for nucleic acid array detection. In the subject invention, a set of a representational number of distinct gene specific primers is used to generate a sub-population of labeled nucleic acids from each of the different physiological samples. The labeled nucleic acids are then compared to each other by hybridizing to a specially designed nucleic acid array that contains the representational number of distinct genes. The subject methods find use in identifying the expression pattern of the genes of special interest in the physiological samples.