Abstract: Suspension bands, helmet systems, and/or the like are provided. In some embodiments, a suspension band includes: a plurality of knitted loop structures; and a plurality of knitted inlay structures, wherein the plurality of knitted inlay structures is inserted as a weft inlay stitch on, beneath, or within the plurality of knitted loop structures. In some embodiments, the suspension band includes one or more inlay sections and a reinforcement section; a greater portion of the plurality of knitted loop structures are disposed on, beneath, or within the reinforcement section than are disposed on, beneath, or within the one or more inlay sections. In some embodiments, the one or more inlay sections include a first inlay section and a second inlay section, wherein the first inlay section is disposed on a first side of the reinforcement section and the second inlay section is disposed on a second side of the reinforcement section.

Abstract: Biosynthetic melanin can capture fentanyl from aqueous environments. The captured fentanyl can be released by introducing the fentanyl-bound melanin to an acidic environment. In this way, fentanyl can capture melanin for subsequent detection/analysis, in addition to operating to remove fentanyl for decontamination.

Abstract: The present disclosure provides a method for producing a LGPS-based solid electrolyte at a significantly lower temperature and/or time compared to conventional methods. In one particular embodiment, methods disclosed herein include subjecting a material comprising an LGPS-based powder to a pressure of at least 250 MPa at a temperature less than about 450° C. under conditions sufficient to produce the LGPS-based solid electrolyte.

Abstract: A method includes receiving a set of training utterances each including a non-synthetic speech representation of a corresponding utterance, and for each training utterance, generating a corresponding synthetic speech representation by using a voice conversion model. The non-synthetic speech representation and the synthetic speech representation form a corresponding training utterance pair. At each of a plurality of output steps for each training utterance pair, the method also includes generating, for output by a speech recognition model, a first probability distribution over possible non-synthetic speech recognition hypotheses for the non-synthetic speech representation and a second probability distribution over possible synthetic speech recognition hypotheses for the synthetic speech representation.

Abstract: A method and system for polymorphic pruning of neural networks are disclosed. The method involves training a neural network model on an input dataset for an initial predetermined number of iterations to gather weight information, including the strength of each weight and changes in strength over iterations. The weights are stored in an array accessible to a pruning algorithm. An objective function is compiled using the weight information, and an optimization tool solves the objective function to generate a solution vector. This solution vector is used to create a pruning mask, which is applied to the neural network model to prune certain weights by setting them to zero. The pruned weight vector updates the model, resulting in a neural network with fewer non-zero connections between neurons.

Abstract: Embodiments of the disclosure provide a scanning mirror assembly. In certain configurations, the scanning mirror assembly may include a two-dimensional micro-electromechanical system (MEMS) scanning mirror, a first pair of piezoelectric electrodes coupled to the MEMS scanning mirror through a first pair of looped torsion springs, and a second pair of piezoelectric electrodes coupled to the MEMS scanning mirror through a second pair of looped torsion springs. The first pair of piezoelectric electrodes drives the MEMS scanning mirror to rotate around a first axis. The second pair of piezoelectric electrodes drives the MEMS scanning mirror to rotate around a second axis orthogonal to the first axis.

Abstract: Often when there is a glare on a display screen the user may be able to mitigate the glare by tilting or otherwise moving the screen or changing their viewing position. However, when driving a car there are limited options for overcoming glares on the dashboard, especially when you are driving for a long distance in the same direction. Embodiments are directed to eliminating such glare. Other embodiments are related to mixed reality (MR) and filling in occluded areas.

Abstract: The subject disclosure relates to storage. An example method includes receiving an input/output (I/O) request based on a protocol related to a first type of storage medium. The method further includes determining a namespace among a plurality of namespaces that is associated with the I/O request. The method further includes determining a binary file corresponding to the determined namespace based on a mapping relationship. The method further includes reading from and/or writing to a second type of storage medium based on the determined binary file, the second type being different from the first type. In this manner, a unified underlying storage implementation with software definition capabilities is provided, which enables different types of storage media to be operated as namespaces, thereby enabling the deployment of a storage system to adapt to more platforms and scenarios.

Abstract: Embodiments include a method of forming a contact structure on a semiconductor substrate. The method including selectively depositing a metal silicide layer over a contact formed within a cavity of a substrate and a bottom surface of the cavity using a selective deposition process, including forming a residual layer on a surface of a dielectric layer forming sidewalls of the cavity, wherein a thickness of the metal silicide layer deposited over the contact is greater than a thickness of the residual layer, removing at least a portion of the residual layer formed on the dielectric layer using an etching process that comprises exposing the metal selectively deposited layer to a metal halide containing precursor, and selectively depositing a metal fill over the metal silicide layer remaining over the contact after removing the at least the portion of the residual layer using a selective metal fill process.

Abstract: Embodiments include a method of forming a contact structure on a semiconductor substrate. The method including selectively depositing a metal silicide layer over a contact formed within a cavity of a substrate and a bottom surface of the cavity using a selective deposition process, including forming a residual layer on a surface of a dielectric layer forming sidewalls of the cavity, wherein a thickness of the metal silicide layer deposited over the contact is greater than a thickness of the residual layer, removing at least a portion of the residual layer formed on the dielectric layer using an etching process that comprises exposing the metal selectively deposited layer to a metal halide containing precursor, and selectively depositing a metal fill over the metal silicide layer remaining over the contact after removing the at least the portion of the residual layer using a selective metal fill process.

Abstract: Techniques for predicting analyte levels based on intermittent sensor data are disclosed. First sensor data is acquired from an analyte sensor. The first sensor data reflects analyte levels of a user who is wearing the analyte sensor. The first sensor data is collected over a first time period. Later, a determination is made as to whether data is still being acquired from the analyte sensor. As a result of determining that data is no longer being acquired from the analyte sensor, the first sensor data is classified as intermittent analyte data. The intermittent analyte data is then used to generate predicted analyte level data for the user during a second time period that is subsequent to the first time period. The predicted analyte level data is reflective of the intermittent analyte data.

Abstract: Methods of manufacturing composite workpieces that include adding an expandable element to an internal volume of a constraining container proximate to a uncured composite workpiece supported on a rigid form, where the expandable element is configured to expand when a predetermined change is produced in an attribute of the expandable element; expanding the expandable element by producing the predetermined change in the attribute of the expandable element, so that an expansion of the expandable element applies pressure to the workpiece supported on the rigid form within the internal volume, and curing the composite workpiece while the resulting pressure is applied to the workpiece supported on the rigid form.

Abstract: A wide-coverage edge ring configured to be arranged above a bottom ring in a substrate processing chamber includes an upper surface, a lower surface that includes a lower surface step that extends downward from the lower surface and is configured to be received within and interface with a pocket defined at least partially by an upper surface of the bottom ring and an inner surface of a chamber liner, an inner diameter, a ledge defined in the inner diameter of the edge ring, and an outer diameter. The outer diameter of the edge ring includes a projection that extends radially outward from the edge ring and defines an inward step in the outer diameter, the projection and the inward step are configured to interface with an upper end of the chamber liner, and the projection is configured to extend at least partially over the upper end of the chamber liner.

Abstract: A method includes receiving a set of training utterances each including a non-synthetic speech representation of a corresponding utterance, and for each training utterance, generating a corresponding synthetic speech representation by using a voice conversion model. The non-synthetic speech representation and the synthetic speech representation form a corresponding training utterance pair. At each of a plurality of output steps for each training utterance pair, the method also includes generating, for output by a speech recognition model, a first probability distribution over possible non-synthetic speech recognition hypotheses for the non-synthetic speech representation and a second probability distribution over possible synthetic speech recognition hypotheses for the synthetic speech representation.

Abstract: A method includes receiving training data including transcribed speech utterances spoken in a general domain, modified speech utterances in a target domain, and unspoken textual utterances corresponding to the transcriptions of the modified speech utterances in the target domain. The modified speech utterances include utterances spoken in the target domain that have been modified to obfuscate one or more classes of sensitive information recited in the utterances. The method also includes generating a corresponding alignment output for each unspoken textual utterance of the received training data using an alignment model. The method also includes training a speech recognition model on the alignment outputs generated for the corresponding to the unspoken textual utterances, the un-transcribed speech utterances, and the transcribed speech utterances to teach the speech recognition model to learn to recognize speech in the target domain and phrases within the one or more classes of sensitive information.

Abstract: Generally described, aspects of the disclosed subject matter are directed to impact mitigating membranes. In accordance with aspects of the present disclosure, the impact mitigating membranes generally include a laminar core having a first lamina and a second lamina coupled to the first lamina and configured to shift relative to the first lamina to dissipate kinetic energy resulting from an oblique impact force. The laminar core may include a lubricating layer or self-lubrication between the lamina to reduce friction. In general, the laminar core is surrounded by first and second encapsulating layers positioned adjacent each lamina. An anchoring attachment coupled to one or more of the first and second encapsulating layer and associating with the laminar core constrains relative lateral motion between the laminar core and the first and second encapsulating layers.

Abstract: A method includes receiving a set of training utterances each including a non-synthetic speech representation of a corresponding utterance, and for each training utterance, generating a corresponding synthetic speech representation by using a voice conversion model. The non-synthetic speech representation and the synthetic speech representation form a corresponding training utterance pair. At each of a plurality of output steps for each training utterance pair, the method also includes generating, for output by a speech recognition model, a first probability distribution over possible non-synthetic speech recognition hypotheses for the non-synthetic speech representation and a second probability distribution over possible synthetic speech recognition hypotheses for the synthetic speech representation.

Abstract: The present disclosure seeks to reduce the effects of rotational and linear acceleration experienced by the body of a user in response to an impact force. Modular disengaging systems of the present disclosure are generally suitable for coupling to protective equipment to provide a disengaging motion between various layers such that the effects of the impact force to the body of the user are reduced. Generally described, the modular disengaging systems of the present disclosure include layers configured to facilitate relative lateral motion therebetween upon an impact force.

Abstract: A method includes receiving training data that includes unspoken text utterances, un-transcribed non-synthetic speech utterances, and transcribed non-synthetic speech utterances. Each unspoken text utterance is not paired with any corresponding spoken utterance of non-synthetic speech. Each un-transcribed non-synthetic speech utterance is not paired with a corresponding transcription. Each transcribed non-synthetic speech utterance is paired with a corresponding transcription. The method also includes generating a corresponding synthetic speech representation for each unspoken textual utterance of the received training data using a text-to-speech model. The method also includes pre-training an audio encoder on the synthetic speech representations generated for the unspoken textual utterances, the un-transcribed non-synthetic speech utterances, and the transcribed non-synthetic speech utterances to teach the audio encoder to jointly learn shared speech and text representations.

Abstract: A method includes receiving audio data corresponding to an utterance and generating a pair of positive audio data examples. Here, each positive audio data example includes a respective augmented copy of the received audio data. For each respective positive audio data example, the method includes generating a respective sequence of encoder outputs and projecting the respective sequence of encoder outputs for the positive data example into a contrastive loss space. The method also includes determining a L2 distance between each corresponding encoder output in the projected sequences of encoder outputs for the positive audio data examples and determining a per-utterance consistency loss by averaging the L2 distances. The method also includes generating corresponding speech recognition results for each respective positive audio data example. The method also includes updating parameters of the speech recognition model based on a respective supervised loss term and the per-utterance consistency loss.