Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scheduling operations represented as a computational graph on a distributed computing network. A method includes: receiving data representing operations to be executed in order to perform a job on a plurality of hardware accelerators of a plurality of different accelerator types; generating, for the job and from at least the data representing the operations, features that represent a predicted performance for the job on hardware accelerators of the plurality of different accelerator types; generating, from the features, a respective predicted performance metric for the job for each of the plurality of different accelerator types according to a performance objective function; and providing, to a scheduling system, one or more recommendations for scheduling the job on one or more recommended types of hardware accelerators.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scheduling operations represented as a computational graph on a distributed computing network. A method includes: receiving data representing operations to be executed in order to perform a job on a plurality of hardware accelerators of a plurality of different accelerator types; generating, for the job and from at least the data representing the operations, features that represent a predicted performance for the job on hardware accelerators of the plurality of different accelerator types; generating, from the features, a respective predicted performance metric for the job for each of the plurality of different accelerator types according to a performance objective function; and providing, to a scheduling system, one or more recommendations for scheduling the job on one or more recommended types of hardware accelerators.

Abstract: In one embodiment, a device in a network receives a machine learning encoder and decoder trained by a supervisory service. The service trains the encoder and decoder using vibration measurement data sent to the service by a plurality of devices. The device trains, based on the received encoder, a classifier to determine whether vibration measurement data is indicative of a behavioral anomaly. The device receives vibration measurement data captured by a particular set of one or more vibration sensors of a monitored system. The device evaluates, using the trained decoder, the received vibration measurement data to determine whether the data is indicative of a structural anomaly in the monitored system. The device evaluates, using the trained classifier, the received vibration measurement data to determine whether the data is indicative of a behavioral anomaly in the monitored system.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scheduling operations represented as a computational graph on a distributed computing network. A method includes: receiving data representing operations to be executed in order to perform a job on a plurality of hardware accelerators of a plurality of different accelerator types; generating, for the job and from at least the data representing the operations, features that represent a predicted performance for the job on hardware accelerators of the plurality of different accelerator types; generating, from the features, a respective predicted performance metric for the job for each of the plurality of different accelerator types according to a performance objective function; and providing, to a scheduling system, one or more recommendations for scheduling the job on one or more recommended types of hardware accelerators.

Abstract: The invention provides a retrofit LED lamp driver having a shunt switch to alternately shunt (for a first duration) and not shunt (for a second duration) received AC power. The first duration is concurring with a first polarity of each AC cycle of the AC power, for a first period comprising at least one AC cycle, and the first duration is concurring in a second, opposite polarity of the AC power, for a second period comprising another at least one AC cycle not overlapping with the first duration. Alternating the first and second periods are provided. Any DC offset in the current drawn caused by the shunting operation is thereby cancelled over time.

Abstract: The invention provides a retrofit LED lamp driver having a shunt switch to alternately shunt (for a first duration) and not shunt (for a second duration) received AC power. The first duration is concurring with a first polarity of each AC cycle of the AC power, for a first period comprising at least one AC cycle, and the first duration is concurring in a second, opposite polarity of the AC power, for a second period comprising another at least one AC cycle not overlapping with the first duration. Alternating the first and second periods are provided. Any DC offset in the current drawn caused by the shunting operation is thereby cancelled over time.

Abstract: In one embodiment, a device in a network receives a machine learning encoder and decoder trained by a supervisory service. The service trains the encoder and decoder using vibration measurement data sent to the service by a plurality of devices. The device trains, based on the received encoder, a classifier to determine whether vibration measurement data is indicative of a behavioral anomaly. The device receives vibration measurement data captured by a particular set of one or more vibration sensors of a monitored system. The device evaluates, using the trained decoder, the received vibration measurement data to determine whether the data is indicative of a structural anomaly in the monitored system. The device evaluates, using the trained classifier, the received vibration measurement data to determine whether the data is indicative of a behavioral anomaly in the monitored system.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for scheduling operations represented as a computational graph on a distributed computing network. A method includes: receiving data representing operations to be executed in order to perform a job on a plurality of hardware accelerators of a plurality of different accelerator types; generating, for the job and from at least the data representing the operations, features that represent a predicted performance for the job on hardware accelerators of the plurality of different accelerator types; generating, from the features, a respective predicted performance metric for the job for each of the plurality of different accelerator types according to a performance objective function; and providing, to a scheduling system, one or more recommendations for scheduling the job on one or more recommended types of hardware accelerators.

Abstract: A programming device is provided that programs cards, such as payment cards, with data, such as personal data, using light transmitters and receivers. For example, an infrared transmitter may be provided to program personal data (e.g., a customer's credit card number) into a card wirelessly. In doing so, the card may be, for example, completely laminated such that there are no exposed electronic components on the exterior surface of the card and be programmed via light. The programming device may shield the programming components to block ambient light from interacting with those programming components during programming. A conveyor may be utilized to align multiple cards with a programming device to allow assembly-line style programming of the cards.

Abstract: Transdermal delivery peptides for the treatment of skin diseases and/or facilitation or enhancement of transdermal delivery of pharmaceutically active agents are provided. Compositions comprising the transdermal delivery peptides and methods of therapeutic use, including the improvement of transdermal delivery of drugs or other pharmaceutically active agents, are also disclosed. Nucleic acids, expression vectors, and methods of their use, which encode the transdermal delivery peptides are disclosed. Methods are also provided for in vivo phage display for identifying further peptides with enhanced transdermal delivery capability.

Abstract: Transdermal delivery peptides for the treatment of skin diseases and/or facilitation or enhancement of transdermal delivery of pharmaceutically active agents are provided. Compositions comprising the transdermal delivery peptides and methods of therapeutic use, including the improvement of transdermal delivery of drugs or other pharmaceutically active agents, are also disclosed. Nucleic acids, expression vectors, and methods of their use, which encode the transdermal delivery peptides are disclosed. Methods are also provided for in vivo phage display for identifying further peptides with enhanced transdermal delivery capability.

Abstract: A bar code scanner having means for reading bar code data and for processing the data includes means for supplying the processed data to a computer having an input/output (I/O) port designed for the connection thereto of an external keyboard. The scanner includes "communication" means for enabling the scanner either to be "wedged" between an external keyboard and the I/O port or to be directly connected to the I/O port. The scanner's "communication" means includes means for sensing whether an external keyboard is connected to the computer and includes means responsive to sensing the absence of an external keyboard connection for producing signals similar to those produced by an external keyboard and for supplying these signals to the computer in order to establish and maintain communication between the scanner and the computer. The bar code scanner may also include means for responding to various computer commands and/or keyboard commands.

Abstract: An automatic electronically triggered scanner includes photo sensing apparatus for reading bar code data and for decoding the bar code data and then ascertaining whether the bar code data is valid. Following the reading and decoding of a valid bar code, the further decoding of bar code data by the scanner is inhibited until a predetermined condition is met to ensure that the bar code just read (which provided valid data) is not re-read. For the predetermined condition to be met the scanner must sense the absence of potentially valid bar code data for a preset interval following the decoding of each set of valid data. Subseqent to the preset interval during which only invalid data is present the scanner is placed in a condition to process valid data.