Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating a binding prediction neural network. In one aspect, a method comprises: instantiating a plurality of structure prediction neural networks, wherein each structure prediction neural network has a respective neural network architecture and is configured to process data defining an input polynucleotide to generate data defining a predicted structure of the input polynucleotide; training each of the plurality of structure prediction neural networks; after training the plurality of structure prediction neural networks, determining a respective performance measure of each structure prediction neural network based at least in part on a prediction accuracy of the structure prediction neural network; and generating, based on the performance measures of the structure prediction neural networks, a binding prediction neural network.

Abstract: A method and apparatus for converting an analog waveform into digital messages and for reconstructing the analog waveform from the information contained in the digital messages. The digital messages describe a change in amplitude of the analog waveform, the polarity or direction of the change in amplitude and the elapsed time for the change in amplitude. Each digital message is created upon the occurrence of a change in amplitude which exceeds a predetermined or predefined threshold voltage. The digital messages may in some instances be combined to compress or lessen the total number of digital messages necessary to reconstruct the analog wave. Using the information in the digital messages, the analog waveform can be reconstructed without the use of smoothing filters.

Abstract: A method and apparatus for converting an analog waveform into digital messages and for reconstructing the analog waveform from the information contained in the digital messages. The digital messages describe a change in amplitude of the analog waveform, the polarity or direction of the change in amplitude and the elapsed time for the change in amplitude. Each digital message is created upon the occurrence of a change in amplitude which exceeds a predetermined or predefined threshold voltage. The digital messages may in some instances be combined to compress or lessen the total number of digital messages necessary to reconstruct the analog wave. Using the information in the digital messages, the analog waveform can be reconstructed without the use of smoothing filters.

Abstract: A media-on-demand system employs predictive memory caching for a media server and associated data storage device to meet requests from multiple clients for media programs stored on the data storage device. The system allocates a plurality of sections of the server memory buffer to accomodate requests for popular titles. When requests for a title are received from a first grouping of clients, the server scheduler establishes a first streaming data process for the requested media program by reading in the data blocks for the program and retaining the data blocks in memory for use by another group of clients if requested close enough in time to the first grouping such that the first data block of the media program is still retained in memory. Used data blocks are removed if no requests by another group of clients are made within a given time and the memory is full or if the title is of a low priority level.

Abstract: A scale incorporating an abuse detection and recordal system has a scale for weighing an article placed thereon, the scale defining a surface adapted to support the article thereon and a component damageable by an excessive force downwardly directed on the support surface. A frangible layer is disposed on the support surface and frangible by a downwardly directed force less than the excessive force required to damage the component, whereby the frangible layer acts as an abuse detection and recordal system to detect and record the use of such force.

Abstract: Apparatus for automatically determining the weight of a plurality of articles, each of the articles having a surface portion with a unique element thereon, comprises a plurality of assemblies and a computer. Each assembly includes a surface for supporting an article, a transducer in operative contact with the supporting surface for producing an output signal indicative of the weight of an article placed on the supporting surface, and a sensor in operative relationship with an article on the supporting surface for producing an output signal indicative of the unique element on the surface portion of the article. A computer receives the transducer output signals and the sensor output signals and, for each article on the supporting surfaces, computes the weight of the article based on the transducer output signal and identifies the article based on the sensor output signal. The apparatus is especially useful as an inventory control system for a bar.