Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for compressing and decompressing data. In one aspect, a method comprises: processing data using an encoder neural network to generate a latent representation of the data; processing the latent representation of the data using a hyper-encoder neural network to generate a latent representation of an entropy model; generating an entropy encoded representation of the latent representation of the entropy model; generating an entropy encoded representation of the latent representation of the data using the latent representation of the entropy model; and determining a compressed representation of the data from the entropy encoded representations of: (i) the latent representation of the data and (ii) the latent representation of the entropy model used to entropy encode the latent representation of the data.

Abstract: An analog conditioning circuit and a corresponding method for processing an analog input signal provide a conditioned analog signal for input into an analog processing circuit. The analog conditioning circuit comprises a main signal path between an input for receiving the analog input signal and an output for outputting the conditioned analog signal, wherein the transfer function of the main signal path is constrained by a transfer function requirement associated with the analog processing circuit; and a feedforward signal path comprising a first filtering block configured to attenuate desired frequencies of a first signal derived from the analog input signal to provide a filtered analog signal; wherein the feedforward signal path is configured to input the filtered analog signal into the main signal path such that the filtered analog signal is subtracted from a second signal derived from the analog input signal to provide the conditioned analog signal.

Abstract: A system, method and non-transitory computer readable medium for labeling a plurality of messages from a customer-agent interaction on a social media service to identify an issue and a response are disclosed. For example, the system includes a conversation interface, a conversation database coupled to the conversation interface, a conversation analysis server coupled to the conversation database and a conversation knowledge repository coupled to the conversation analysis server. The conversation analysis server includes a preprocessing module, a dialogue act analysis module, an issue status analysis module and an issue/response identification module.

Abstract: A system may include an output stage for driving a load at an output of the output stage, a pulse-width modulation mode path configured to pre-drive the output stage in a first mode of operation, a linear mode path configured to pre-drive the output stage in a second mode of operation and a loop filter coupled at its input to the output of the output stage and coupled at its output to both of the pulse-width modulation mode path and the linear mode path.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for compressing and decompressing data. In one aspect, an encoder neural network processes data to generate an output including a representation of the data as an ordered collection of code symbols. The ordered collection of code symbols is entropy encoded using one or more code symbol probability distributions. A compressed representation of the data is determined based on the entropy encoded representation of the collection of code symbols and data indicating the code symbol probability distributions used to entropy encode the collection of code symbols. In another aspect, a compressed representation of the data is decoded to determine the collection of code symbols representing the data. A reconstruction of the data is determined by processing the collection of code symbols by a decoder neural network.

Abstract: Various arrangements are presented for determining a state of a traffic light. A first analysis of a scene may be performed that includes the traffic light. The first analysis may result in the output of a first classification result and a first confidence score. A second, different analysis of the scene may be performed that also outputs a classification result and a confidence score. A fusion process may be performed to fuse the classification results into a fused classification result based on the confidence scores and one or more weighting metrics. A vehicle may then be driven at least partially based on the fused classification result.

Abstract: A method of preparing a Ag2ZnSn(S,Se)4 compound, including dissolving selenourea (SeC(NH2)2) in an aprotic solvent, and dissolving a silver salt, a zinc salt, and a tin salt in the aprotic solvent with the selenourea to form a metal solution; and coating the metal solution onto a substrate to form an Ag2ZnSn(S,Se)4 compound layer on the substrate.

Abstract: The present disclosure provides systems and methods that enable distance metric learning using proxies. A machine-learned distance model can be trained in a proxy space in which a loss function compares an embedding provided for an anchor data point of a training dataset to a positive proxy and one or more negative proxies, where each of the positive proxy and the one or more negative proxies serve as a proxy for two or more data points included in the training dataset. Thus, each proxy can approximate a number of data points, enabling faster convergence. According to another aspect, the proxies of the proxy space can themselves be learned parameters, such that the proxies and the model are trained jointly. Thus, the present disclosure enables faster convergence (e.g., reduced training time). The present disclosure provides example experiments which demonstrate a new state of the art on several popular training datasets.

Abstract: Systems and methods for secured access to cloud-based applications or services include a service node that may receive a request from client including a URL associated with an application manager. The service node may send a URL prefix identifying a termination to the termination node. The service node may receive a client hello message from the client that includes a first field incorporating the URL prefix, and may send the client hello message to the termination node to initiate a handshake with the client using a wildcard certificate of server, for establishing a SSL channel between the client and the termination node for a session of the application. The service node can direct a communication of the session from the client to the predetermined termination node, for decryption, using the established SSL channel, according to the URL prefix incorporated in a server name indication (SNI) field of the communication.

Abstract: Low-voltage rechargeable microbatteries are provided. In one aspect, a method of forming a microbattery includes: forming a cathode on a substrate, wherein the cathode includes a lithium intercalated material; forming a solid electrolyte on the cathode; forming an anode on the solid electrolyte; and forming a negative contact on the anode. A microbattery is also provided.

Abstract: Low-voltage rechargeable microbatteries having a vanadium-based cathode are provided. In one aspect, a method of forming a battery is provided. The method includes the steps of: forming a first contact on a substrate; forming a cathode on the first contact, wherein the cathode is formed from a vanadium-containing material; forming a solid electrolyte on the cathode; forming an anode on the solid electrolyte; and forming a second contact on the anode. A battery having a vanadium-based cathode is also provided.

Abstract: A computer-implemented method for transferring reward points between mobile computing devices is implemented by a first mobile computing device. The method includes receiving a first reward points balance associated with a first reward points account, detecting a second mobile computing device associated with a second reward points account wherein the second reward points account has a second reward points balance, initiating a transfer request from the first mobile computing device to the second mobile computing device, and sending a transfer command to a rewards program computer system wherein the transfer command instructs the rewards program computer system to transfer the at least one reward point between the first reward points account and the second reward points account.

Abstract: A method for generating a filename for a chunk of streamed video content is disclosed. The method comprises performing a cryptographic hash function on data associated with the chunk of video content (100A) and setting the output hash value of the cryptographic hash function as the chunk filename (100B). A method for recording broadcast video content by a user device is also disclosed. The method comprises receiving a chunk of broadcast video content (310), generating a filename for the received chunk of video content (320) and storing the generated filename in a user specific storage (330). A method for broadcasting a chunk of video content is also disclosed. The method comprises generating metadata corresponding to the chunk of video content (550), the metadata comprising at least one parameter for generating a file name for the chunk of video content, and broadcasting the generated metadata with the chunk of video content (560).

Abstract: In accordance with embodiments of the present disclosure, a processing system may include a plurality of processing paths and a controller. The plurality of processing paths may include a static processing path configured to generate a first digital signal based on an analog input signal and a dynamic processing path configured to generate a second digital signal based on the analog input signal, wherein a parameter of the dynamic processing path is determined based on a characteristic of the analog input signal. The controller may be configured to select the first digital signal as a digital output signal of the processing system when a change is occurring to the characteristic and select the second digital signal as the digital output signal in the absence of change occurring to the characteristic.

Abstract: A method may include, in an apparatus comprising a closed loop amplifier and a signal processing block configured to generate an amplifier input signal as a function of an upstream signal received at an input of the signal processing block, in a calibration mode of the apparatus: decoupling a second stage input of the amplifier from a first stage output of the amplifier; determining an offset signal that when applied to the input of a signal processing block as the upstream signal generates approximately zero as an intermediate signal generated by the first stage of the amplifier; and controlling one or more parameters of the apparatus based on the offset signal to compensate for an offset of at least one of the first stage and the signal processing block.

Abstract: The present disclosure provides systems and methods that enable distance metric learning using proxies. A machine-learned distance model can be trained in a proxy space in which a loss function compares an embedding provided for an anchor data point of a training dataset to a positive proxy and one or more negative proxies, where each of the positive proxy and the one or more negative proxies serve as a proxy for two or more data points included in the training dataset. Thus, each proxy can approximate a number of data points, enabling faster convergence. According to another aspect, the proxies of the proxy space can themselves be learned parameters, such that the proxies and the model are trained jointly. Thus, the present disclosure enables faster convergence (e.g., reduced training time). The present disclosure provides example experiments which demonstrate a new state of the art on several popular training datasets.

Abstract: The present disclosure provides systems and methods that enable distance metric learning using proxies. A machine-learned distance model can be trained in a proxy space in which a loss function compares an embedding provided for an anchor data point of a training dataset to a positive proxy and one or more negative proxies, where each of the positive proxy and the one or more negative proxies serve as a proxy for two or more data points included in the training dataset. Thus, each proxy can approximate a number of data points, enabling faster convergence. According to another aspect, the proxies of the proxy space can themselves be learned parameters, such that the proxies and the model are trained jointly. Thus, the present disclosure enables faster convergence (e.g., reduced training time). The present disclosure provides example experiments which demonstrate a new state of the art on several popular training datasets.

Abstract: Low temperature techniques for forming layered lithium cobalt oxide (LCO) are provided. In one aspect, a method of synthesizing layered LCO includes: forming a metal catalyst layer (e.g., platinum) on a substrate; depositing LCO onto the metal catalyst layer; and annealing the LCO under conditions sufficient to form the layered LCO on the metal catalyst layer. An adhesion layer can be deposited on the substrate, and the metal catalyst layer can be deposited onto the adhesion layer. In another aspect, a structure is provided including: a substrate; a metal catalyst layer (e.g., platinum) disposed on the substrate; and layered LCO formed on the metal catalyst layer. An adhesion layer can be disposed between the substrate and the metal catalyst layer.

Abstract: A method of preparing a Ag2ZnSn(S,Se)4 compound, including dissolving selenourea (SeC(NH2)2) in an aprotic solvent, and dissolving a silver salt, a zinc salt, and a tin salt in the aprotic solvent with the selenourea to form a metal solution; and coating the metal solution onto a substrate to form an Ag2ZnSn(S,Se)4 compound layer on the substrate.

Abstract: Embodiments are directed to a memristive device. The memristive device includes a first conductive material layer. An oxide material layer is arranged on the first conductive layer. And a second conductive material layer is arranged on the oxide material layer, wherein the second conductive material layer comprises a metal-alkali alloy.