Christian Szegedy

Bio: Christian Szegedy is an academic researcher from Google. The author has contributed to research in topics: Automated theorem proving & Deep learning. The author has an hindex of 31, co-authored 69 publications receiving 147148 citations. Previous affiliations of Christian Szegedy include Lawrence Berkeley National Laboratory & Cadence Design Systems.

Business discovery from imagery

Abstract: Aspects of the present disclosure relate to a method includes training a deep neural network using training images and data identifying one or more business storefront locations in the training images The deep neural network outputs tight bounding boxes on each image At the deep neural network, a first image may be received The first image may be evaluated using the deep neural network Bounding boxes may then be generated identifying business storefront locations in the first image

Method, system, and computer program product for implementing incremental placement in electronics design

Abstract: Disclosed are a method, system, and computer program product for implementing incremental placement for an electronic design while predicting and minimizing the perturbation impact arising from incremental placement of electronic components. In some embodiments, an initial placement of an electronic design is identified, the abstract flow is computed, the target locations of various electronic components to be placed are identified, the relative ordering of electronic components are determined, and the placement is then legalized. Furthermore, in various embodiments, the method, system, or computer program product starts with an initial placement of an electronic design and derives a legal placement by using the incremental placement technique while minimizing the perturbation impact or the total quadratic movement of instances. In some embodiments, an augmented or incremental clumping technique based data structure is utilized for rapid and substantially exact perturbation prediction of effects of local incremental placement operations.

Methods, systems, and computer program products for grid-morphing techniques in placement, floorplanning, and legalization

Abstract: Disclosed are methods, systems, and computer program products for performing grid morphing technique for computing a spreading of objects over an area such that the final locations of the objects are distributed over the area and such that the final locations of the objects are minimally perturbed from their initial starting locations and the density of objects meets certain constraints. The minimization of perturbation, or stability, of the approaches disclosed, is the key feature which is the principal benefit of the techniques disclosed. The methods described herein may be used as part of a tool for placement or floorplanning of logic gates or larger macroblocks for the design of an integrated circuit.

Systems and Methods for Identifying Entities Directly from Imagery

Abstract: Systems and methods of identifying entities are disclosed. In particular, one or more images that depict an entity can be identified from a plurality of images. One or more candidate entity profiles can be determined from an entity directory based at least in part on the one or more images that depict the entity. The one or more images that depict the entity and the one or more candidate entity profiles can be provided as input to a machine learning model. One or more outputs of the machine learning model can be generated. Each output can include a match score associated with an image that depicts the entity and at least one candidate entity profile. The entity directory can be updated based at least in part on the one or more generated outputs of the machine learning model.

Symplectic Spaces And Ear-Decomposition Of Matroids

Abstract: Matroids admitting an odd ear-decomposition can be viewed as natural generalizations of factor-critical graphs. We prove that a matroid representable over a field of characteristic 2 admits an odd ear-decomposition if and only if it can be represented by some space on which the induced scalar product is a non-degenerate symplectic form. We also show that, for a matroid representable over a field of characteristic 2, the independent sets whose contraction admits an odd ear-decomposition form the family of feasible sets of a representable Δ-matroid.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Very Deep Convolutional Networks for Large-Scale Image Recognition

Abstract: In this work we investigate the effect of the convolutional network depth on its accuracy in the large-scale image recognition setting. Our main contribution is a thorough evaluation of networks of increasing depth using an architecture with very small (3x3) convolution filters, which shows that a significant improvement on the prior-art configurations can be achieved by pushing the depth to 16-19 weight layers. These findings were the basis of our ImageNet Challenge 2014 submission, where our team secured the first and the second places in the localisation and classification tracks respectively. We also show that our representations generalise well to other datasets, where they achieve state-of-the-art results. We have made our two best-performing ConvNet models publicly available to facilitate further research on the use of deep visual representations in computer vision.

Deep Residual Learning for Image Recognition

Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers---8x deeper than VGG nets but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.

Going deeper with convolutions

Abstract: We propose a deep convolutional neural network architecture codenamed Inception that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14). The main hallmark of this architecture is the improved utilization of the computing resources inside the network. By a carefully crafted design, we increased the depth and width of the network while keeping the computational budget constant. To optimize quality, the architectural decisions were based on the Hebbian principle and the intuition of multi-scale processing. One particular incarnation used in our submission for ILSVRC14 is called GoogLeNet, a 22 layers deep network, the quality of which is assessed in the context of classification and detection.