Abstract: Techniques for predicting an answer to a question using a machine learning model are described. In some examples, the model predicts one or more answers to the question by: predicting at least two answers to the question using a first component of the question-answer model from a set of passages, generating, using a second component of the question-answer model, at least one question for each of the predicted at least two answers, and performing roundtrip predictions until each generated question only has one answer.

Abstract: Systems and methods for energy management and demand-based installation of charging stations are described. In one aspect, a modular charger unit includes a housing configured to receive cartridges; an interconnect board with slots configured to receive boards housed in cartridges; a computational board housed in a cartridge removably inserted to a first opening of the housing and to a first slot of the interconnect board; a variable high voltage output board housed in a cartridge removably inserted to a second opening of the housing and to a second slot of the interconnect board; a power board housed in a cartridge removably inserted to a third opening of the housing and to a third slot of the interconnect board; and an optional board housed in a cartridge removably inserted to a fourth opening of the housing and to a fourth slot of the interconnect board.

Abstract: According to aspects of the disclosure, a display system is provided comprising a support, a first display coupled to the support, wherein the first display is configured to display output information received from at least one computing device, and a second display rotatably coupled to the first display and rotatably coupled to the support, wherein the second display is configured to display output information received from the at least one computing device.

Abstract: A method of repairing oxidatively-treated hair, the methods comprising the sequential steps of: (i) soaking hair in an aqueous treatment composition, (ii) rinsing the hair of step (i); characterised in that the aqueous treatment composition comprises 0.1 to 6 wt % N-acetyl amino acid, based on the total weight of the composition, wherein the N-acetyl amino acid is selected from the group consisting of N-acetyl glycine, N-acetyl alanine, N-acetyl proline, N-acetyl leucine, N-acetyl isoleucine, N-acetyl valine, N-acetyl tyrosine, N-acetyl phenylalanine, N-acetyl tryptophan, N-acetyl serine, N-acetyl glutamine, N-acetyl threonine, N-acetyl asparagine, N-acetyl methionine, N-acetyl lysine, N-acetyl histidine and mixtures thereof, gives damage repair benefits.

Abstract: Disclosed herein are pyrazole pyrimidine compounds that modulate and/or inhibit hematopoietic progenitor kinase 1, as well as methods of making such compounds and therapeutic methods of using same.

Abstract: Embodiments of a production-quality text-to-speech (TTS) system constructed from deep neural networks are described. System embodiments comprise five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For embodiments of the segmentation model, phoneme boundary detection was performed with deep neural networks using Connectionist Temporal Classification (CTC) loss. For embodiments of the audio synthesis model, a variant of WaveNet was created that requires fewer parameters and trains faster than the original. By using a neural network for each component, system embodiments are simpler and more flexible than traditional TTS systems, where each component requires laborious feature engineering and extensive domain expertise. Inference with system embodiments may be performed faster than real time.

Abstract: Presented herein are embodiments of state-of-the-art speech recognition systems developed using end-to-end deep learning. In embodiments, the model architecture is significantly simpler than traditional speech systems, which rely on laboriously engineered processing pipelines; these traditional systems also tend to perform poorly when used in noisy environments. In contrast, embodiments of the system do not need hand-designed components to model background noise, reverberation, or speaker variation, but instead directly learn a function that is robust to such effects. Neither a phoneme dictionary, nor even the concept of a “phoneme,” is needed. Embodiments include a well-optimized recurrent neural network (RNN) training system that can use multiple GPUs, as well as a set of novel data synthesis techniques that allows for a large amount of varied data for training to be efficiently obtained.

Abstract: Disclosed herein is a polymeric film, the film comprising a polymeric matrix material, a plurality of perovskite nanocrystals and/or aggregates of perovskite nanocrystals dispersed throughout the polymeric matrix material. There is also disclosed a perovskite polymer resin composition, a perovskite-polymer resin composition, a perovskite ink and a method of forming a luminescent film using any one of the compositions or ink. Preferably, the perovskite material is a lead halide perovskite containing a cation selected from Cs, an alkylammonium ion, or a formamidinium ion. The polymeric matrix is preferably formed from monomers comprising a vinyl or an acrylate group.

Abstract: Disclosed herein are pyrazole pyrimidine compounds that modulate and/or inhibit hematopoietic progenitor kinase 1, as well as methods of making such compounds and therapeutic methods of using same.

Abstract: Provided are molecular feedback circuits employing caged-degrons. Aspects of such circuits include the use of a caged-degron to modulate the output of a signaling pathway in a feedback-controlled manner. Also provided are nucleic acids encoding molecular circuits and cells containing such nucleic acids. Methods of using caged-degron-based molecular feedback circuits are also provided, including e.g., methods of modulating a signaling pathway of a cell that include genetically modifying the cell with a caged-degron-based molecular feedback circuit.

Abstract: Provided are molecular feedback circuits as well as nucleic acids encoding such molecular feedback circuits and cells genetically modified with the subject molecular feedback circuits. Methods of modulating signaling of a signaling pathway of a cell using molecular feedback circuits and methods of treating a subject for a condition by administering a cell containing a nucleic acid that encodes a molecular feedback circuit are also provided. Aspects of the molecular feedback circuits of the present disclosure include a signaling protein, of a signaling pathway, that includes a latent deactivation domain. Such circuits may include a regulatory sequence that is responsive to an output of the signaling pathway and is operably linked to a nucleic acid encoding a switch polypeptide that, when expressed, triggers the deactivation domain to deactivate the signaling molecule.

Abstract: Disclosed herein are non-naturally occurring cage polypeptides, kits and degron LOCKRs including the cage polypep-tides, and uses thereof, wherein the cage polypeptides include (a) a helical bundle, comprising between 2 and 7 alpha-helices, wherein the helical bundle includes: (i) a structural region; and (ii) a latch region, wherein the latch region composes a degron located within the latch region, wherein the structural region interacts with the latch region to prevent activity of the degron; and (b) amino acid linkers connecting each alpha helix

Abstract: A method of repairing oxidatively-treated hair, the methods comprising the sequential steps of: (i) soaking hair in an aqueous treatment composition, (ii) rinsing the hair of step (i); characterised in that the aqueous treatment composition comprises 0.1 to 6 wt % N-acetyl amino acid, based on the total weight of the composition, wherein the N-acetyl amino acid is selected from the group consisting of N-acetyl glycine, N-acetyl alanine, N-acetyl proline, N-acetyl leucine, N-acetyl isoleucine, N-acetyl valine, N-acetyl tyrosine, N-acetyl phenylalanine, N-acetyl tryptophan, N-acetyl serine, N-acetyl glutamine, N-acetyl threonine, N-acetyl asparagine, N-acetyl methionine, N-acetyl lysine, N-acetyl histidine and mixtures thereof, gives damage repair benefits.

Abstract: Disclosed herein are pyrozolo-pyrimidin-amino-cycloalkyl compounds, analogs thereof, pharmaceutical compositions comprising thereof and therapeutic uses therefor.

Abstract: Embodiments of a production-quality text-to-speech (TTS) system constructed from deep neural networks are described. System embodiments comprise five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For embodiments of the segmentation model, phoneme boundary detection was performed with deep neural networks using Connectionist Temporal Classification (CTC) loss. For embodiments of the audio synthesis model, a variant of WaveNet was created that requires fewer parameters and trains faster than the original. By using a neural network for each component, system embodiments are simpler and more flexible than traditional TTS systems, where each component requires laborious feature engineering and extensive domain expertise. Inference with system embodiments may be performed faster than real time.

Abstract: Disclosed herein is a polymeric film, the film comprising a polymeric matrix material, a plurality of perovskite nanocrystals and/or aggregates of perovskite nanocrystals dispersed throughout the polymeric matrix material. There is also disclosed a perovskite polymer resin composition, a perovskite-polymer resin composition, a perovskite ink and a method of forming a luminescent film using any one of the compositions or ink. Preferably, the perovskite material is a lead halide perovskite containing a cation selected from Cs, an alkylammonium ion, or a formamidinium ion. The polymeric matrix is preferably formed from monomers comprising a vinyl or an acrylate group.

Abstract: Embodiments of a production-quality text-to-speech (TTS) system constructed from deep neural networks are described. System embodiments comprise five major building blocks: a segmentation model for locating phoneme boundaries, a grapheme-to-phoneme conversion model, a phoneme duration prediction model, a fundamental frequency prediction model, and an audio synthesis model. For embodiments of the segmentation model, phoneme boundary detection was performed with deep neural networks using Connectionist Temporal Classification (CTC) loss. For embodiments of the audio synthesis model, a variant of WaveNet was created that requires fewer parameters and trains faster than the original. By using a neural network for each component, system embodiments are simpler and more flexible than traditional TTS systems, where each component requires laborious feature engineering and extensive domain expertise. Inference with system embodiments may be performed faster than real time.

Abstract: Disclosed herein are pyrozolo-pyrimidin-amino-cycloalkyl compounds, analogs thereof, pharmaceutical compositions comprising thereof and therapeutic uses therefor.

Abstract: Presented herein are embodiments of state-of-the-art speech recognition systems developed using end-to-end deep learning. In embodiments, the model architecture is significantly simpler than traditional speech systems, which rely on laboriously engineered processing pipelines; these traditional systems also tend to perform poorly when used in noisy environments. In contrast, embodiments of the system do not need hand-designed components to model background noise, reverberation, or speaker variation, but instead directly learn a function that is robust to such effects. A phoneme dictionary, nor even the concept of a “phoneme,” is needed. Embodiments include a well-optimized recurrent neural network (RNN) training system that can use multiple GPUs, as well as a set of novel data synthesis techniques that allows for a large amount of varied data for training to be efficiently obtained.