Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a neural network. One of the methods includes receiving a training image and a ground truth super-resolution image; processing a first training network input comprising the training image using the neural network to generate a first training super-resolution image; processing a first critic input generated from (i) the training image and (ii) the ground truth super-resolution image using a critic neural network to map the first critic input to a latent representation; processing a second critic input generated from (i) the training image and (ii) the first training super-resolution image using the critic neural network to map the second critic input to a latent representation; determining a gradient of a generator loss function that measures a distance between the latent representations of the critic inputs; and determining an update to the parameters.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a machine learning model.

Abstract: Methods, systems, and apparatus, including computer programs encoded on computer storage media, for training a neural network. One of the methods includes receiving a training image and a ground truth super-resolution image; processing a first training network input comprising the training image using the neural network to generate a first training super-resolution image; processing a first critic input generated from (i) the training image and (ii) the ground truth super-resolution image using a critic neural network to map the first critic input to a latent representation; processing a second critic input generated from (i) the training image and (ii) the first training super-resolution image using the critic neural network to map the second critic input to a latent representation; determining a gradient of a generator loss function that measures a distance between the latent representations of the critic inputs; and determining an update to the parameters.

Abstract: A web page or group of pages can be automatically generated based on images selected by a user. The images are analyzed for color and groups of colors for the page background and image framing are determined. Images are analyzed to determine the shapes of the images and the images are placed within appropriate shapes on a page on the web. Click-through rate can be estimated for images and this click-through rate used during the web page generation. Important sections are evaluated in the images and cropping may occur for the images on a page. Text and video can be part of the web page. Software applications can be accessed from the web page as well. The online web pages can be tailored for presentation on a mobile device.

Abstract: A web page or group of pages can be automatically generated based on images selected by a user. The images are analyzed for color and groups of colors for the page background and image framing are determined. Images are analyzed to determine the shapes of the images and the images are placed within appropriate shapes on a page on the web. Click-through rate can be estimated for images and this click-through rate used during the web page generation. Important sections are evaluated in the images and cropping may occur for the images on a page. Text and video can be part of the web page. Software applications can be accessed from the web page as well. The online web pages can be tailored for presentation on a mobile device.

Abstract: Techniques in which an optimal set of clock gating elements is determined for a selected circuit design. Those clock gating elements are coupled to selected flip-flops, with the effect that those selected flip-flops will consume less dynamic power during operation of the logic circuit. The selected set of clock gating elements provides an optimal savings in overall power consumption after modification of that selected circuit design.

Abstract: A circuit ? is coupled to an individual node Nin, in a circuit for which repeated logical values of that individual node can be identified as having a set of flip-flops Fj dependent thereon, with the effect that if the individual node Nin remains unchanged for one or more clock cycles, the set of dependent flip-flops Fj can be disabled for the second and succeeding clock cycles. The circuit ? conditionally generates a clock-enabling signal Nout in response thereto. One such circuit ? conditionally includes a logical controller, whose output is coupled using a fan-out node to both an input to a state machine and a fan-in logic circuit (such as an AND gate). The flip-flop is clocked normally; its output is coupled to that same fan-in logic circuit, whose output Nout is coupled to the set of dependent flip-flops Fj.