Google Inc.

Sequence Design, Inc.

Other

Knowledge Media Inc.

Glen Anderson

Michael Schuster

Amin Vahdat

Junlan Zhou

Alok Kumar

Gregory Sean Corrado

Hrishikesh Balkrishna Aradhye

Peter W Hammond

Uday Ramakrishna Naik

Tushar Deepak Chandra

Tal Shaked

Wei Chai

Sushant Jain

Jeremiah Harmsen

Steven Padgett

Mustafa Ispir

Sivasankar Radhakrishnan

Peter Hanson

Rohan Anil

Xiaobing Liu

Mehmet Levent Koc

Heng-Tze Cheng

David S Yotive

Erwin J Littman

A system includes one or more computers and one or more storage devices storing instructions that when executed by the one or more computers cause the computers to implement a combined machine learning model for processing an input including multiple features to generate a predicted output for the machine learning input. The combined model includes: a deep machine learning model configured to process the features to generate a deep model output; a wide machine learning model configured to process the features to generate a wide model output; and a combining layer configured to process the deep model output generated by the deep machine learning model and the wide model output generated by the wide machine learning model to generate the predicted output, in which the deep model and the wide model have been trained jointly on training data to generate the deep model output and the wide model output.

Wide and deep machine learning models

The present disclosure relates to applying techniques similar to those used in neural network language modeling systems to a content recommendation system. For example, by associating consumed media content to words of a language model, the system may provide content predictions based on an ordering. Thus, the systems and techniques described herein may produce enhanced prediction results for recommending content (e.g. word) in a given sequence of consumed content. In addition, the system may account for additional user actions by representing particular actions as punctuation in the language model.

Content recommendation system using a neural network language model

Systems and methods for achieving high utilization of a network link are provided. A first communication protocol can be selected for transmitting network flows of a first type. A first quality of service can be assigned to network flows of the first type. A second communication protocol can be selected for transmitting network flows of a second type. A second quality of service, lower than the first quality of service, can be assigned to network flows of the second type. A first percentage of available bandwidth can be allocated to the network flows of both the first and second types. The remaining bandwidth, plus a second percentage of available bandwidth, can be allocated to the network flows of the second type, such that the total allocated bandwidth exceeds the available bandwidth of the network link.

Systems and methods for achieving high network link utilization

Techniques and mechanisms for evaluating a transmission outcome indicium to determine a state of data traffic congestion in a network. In an embodiment, congestion information for a packet of a transmission group is evaluated at a computer device to determine a congestion state, where a size of the transmission group is determined based on a congestion window parameter. In another embodiment, the evaluation to determine the congestion state is performed prior to the computer device detecting a transmission outcome indicium for some other packet of the transmission group.

Congestion window modification

A method, apparatus and networking equipment for performing flow hashing is provided. The method may include obtaining, at a network node, a key for a packet; forming, at the network node, a first pseudorandom number as a function of a first polynomial generator and the key; forming, at the network node, a second pseudorandom number as a function of a second polynomial generator and the key; combining, at the network node, the first and second pseudorandom numbers to form a third pseudorandom number; and selecting, at the network node, as a function of the third pseudorandom number, a path to which to forward the packet.

Method, apparatus and networking equipment for performing flow hashing using quasi cryptographic hash functions

An embodiment of the present invention is a method for logic synthesis that reduces use of computer memory and reduces computer runtime. In particular, an embodiment of the present invention is a method for logic synthesis which includes the steps of: (a) analyzing an HDL model to develop a parse tree and (b) elaborating the parse tree to create a word-oriented netlist.

Method and apparatus for logic synthesis (word oriented netlist)

An embodiment of the present invention is a method for logic synthesis that reduces use of computer memory and reduces computer runtime. In particular, an embodiment of the present invention is a method for logic synthesis which includes the steps of: (a) analyzing an HDL model to develop a parse tree and (b) elaborating the parse tree to create a word-oriented netlist; wherein elaborating comprises word-oriented elaborating.

Method and apparatus for logic synthesis with elaboration

An embodiment of the present invention is a method for logic synthesis that reduces use of computer memory and reduces computer runtime. In particular, an embodiment of the present invention is a method for logic synthesis which includes the steps of: (a) analyzing an HDL model to develop a parse tree; (b) elaborating the parse tree to create a word-oriented netlist; and (c) inferring complex components from the word-oriented netlist.

Method and apparatus for logic synthesis (inferring complex components)

A method enables an image to be displayed along with annotations in synchronization with separate, time-based program material as might be found in a computer-based multimedia system. The time-based program material may include audio or video information, or a combination of these, as in a digital movie with the annotated image and movie being accessed from a high-capacity storage means, such as a CD-ROM, and played in separate windows on the same display screen. The method provides smooth, real-time annotation of the on-screen image by synchronizing the movement of a pen image with very low flicker and with little overhead in terms of system performance and memory allocation. This improved performance stems from the inventive use of multiple off-screen buffers, and the processing of the previously stored annotations in the form of multiple data records including temporal and spatial information relating to the annotations as originally performed. The data records are processed during predetermined time periods, with image information being transferred into and out of these various off-screen buffers so as to facilitate real-time annotation in synchronization with the time-based program material.

Time-synchronized annotation method

An arc diverter detects an arc between the cathode and the anode of a glow discharge apparatus and in response to arc detection, actuates a switching network that triggers a pair of switches to allow a capacitor to discharge toward the cathode along a resonant diverting path and recharge the capacitor for subsequent discharge in an opposite direction. This discharge current effectively short circuits the power supply and extinguishes the arc. The resonant diverting circuit includes two resonant diverting paths and four switches configured in an 'H' to hold the capacitor in a charged condition. Alternate triggering of the diagonally situated pairs of switches enables the charged capacitor to discharge in one direction along one of the resonant diverting paths and recharge for subsequent discharge in the opposite direction, along the other of the resonant paths. Both resonant diverting paths provide discharge toward the cathode.

Circuit for detecting and diverting an electrical arc in a glow

A prime mover having a central body with a plurality of elongated legs pivotally connected to said central body for movement both up and down and sideways and individual steerable drive means on each elongated leg to enable the prime mover to move over a wide variety of terrains. The drive means include ground supporting wheels or tracks. Carried on the prime mover are electro hydraulic control means for steering each of the drive means and electric control means for actuating each of the drive means for moving the prime mover along the ground. A work platform is supported on the central body, which is appropriately levelled and stabilized in work position by proper extension or retraction or vertical depression or elevation of the respective elongated legs. The prime mover has four independently mounted drive means or wheels, unconnected by common axles or mechanical linkage and having no fixed track, that are controlled in such a manner that the geometry of steering produces a coherent effect.

Growing community of inventors

Palo Alto, CA, United States of America

Glen Anderson