We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overriding in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.

/pdf/imagenet-classification-with-deep-convolutional-neural-1190x9a6b9.pdf

ImageNet Classification with Deep Convolutional Neural Networks

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

The explosion of image data on the Internet has the potential to foster more sophisticated and robust models and algorithms to index, retrieve, organize and interact with images and multimedia data. But exactly how such data can be harnessed and organized remains a critical problem. We introduce here a new database called “ImageNet”, a large-scale ontology of images built upon the backbone of the WordNet structure. ImageNet aims to populate the majority of the 80,000 synsets of WordNet with an average of 500-1000 clean and full resolution images. This will result in tens of millions of annotated images organized by the semantic hierarchy of WordNet. This paper offers a detailed analysis of ImageNet in its current state: 12 subtrees with 5247 synsets and 3.2 million images in total. We show that ImageNet is much larger in scale and diversity and much more accurate than the current image datasets. Constructing such a large-scale database is a challenging task. We describe the data collection scheme with Amazon Mechanical Turk. Lastly, we illustrate the usefulness of ImageNet through three simple applications in object recognition, image classification and automatic object clustering. We hope that the scale, accuracy, diversity and hierarchical structure of ImageNet can offer unparalleled opportunities to researchers in the computer vision community and beyond.

/pdf/imagenet-a-large-scale-hierarchical-image-database-1e8f663s1h.pdf

ImageNet: A large-scale hierarchical image database

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.

/pdf/going-deeper-with-convolutions-1yobw2o2ds.pdf

Going deeper with convolutions

As language models increase in size by the day, methods for efficient inference are critical to leveraging their capabilities for various applications. Prior work has investigated techniques like model pruning, knowledge distillation, and data multiplexing to increase model throughput without sacrificing accuracy. In this paper, we combine two such methods -- structured pruning and data multiplexing -- to compound the speedup gains obtained by either method. Our approach, PruMUX, obtains up to 7.5-29.5X throughput improvement over BERT-base model with accuracy threshold from 80% to 74%. We further study various combinations of parameters (such as sparsity and multiplexing factor) in the two techniques to provide a comprehensive analysis of the tradeoff between accuracy and throughput in the resulting models. We then propose Auto-PruMUX, a meta-level model that can predict the high-performance parameters for pruning and multiplexing given a desired accuracy loss budget, providing a practical method to leverage the combination effectively.

PruMUX: Augmenting Data Multiplexing with Model Compression

To solve the problem of wind noise and structural resistance caused by vehicle external structural parameters, this paper carried out multi-objective optimization for the tail multi-parameters. Based on the selected four parameters of C-Pillar angle, trunk angle, rear box length and boat tail angle of Ahmed model, the central combination test design of response surface Box-Benhnken was carried out in the selected range. The sampled data sets were numerically simulated in star-ccm+, and the surface sound pressure level and drag coefficient of the model were obtained. The optimal Latin hypercube sampling data sets were used to verify the accuracy. Finally, the response surface model was established, and the optimal front window surface SPL was 65.11dB with an error of 1.36%, and the optimal drag coefficient was 0.1210 with an error of 1.68%, taking Curle SPL and drag coefficient as the optimization objective.

Tail multi-parameter optimization of Ahmed model based on response surface method

The structure of the long-span transmission tower is a typical nonlinear structure with the characteristics of great height, large line span, heavy overall weight and flexible tower body. The current design code only analyzes the traditional tower types, but the analysis of the truss structure of transmission tower is limited. Aiming at improving the design defects of the structure of long-span transmission towers, this paper uses the finite element software APDL to build the three-dimensional finite element model of a long-span transmission tower, to carry out the modal finite element analysis as well as to extract the specific parameters of each modal finite element mode: Modality, Natural frequency of vibration, Periodicity. The results show that the natural vibration period of the main machinery of this type of steel transmission tower is about 0.37–1.37 s; The structure of the long-span transmission tower has certain displacements in six degrees of freedom, in which the value of the X-dimensional displacement is the largest. There are some large displacements and local torsion in the high-order mode, combined with the results of modal analysis, so it is suggested to consider the structural improvement or external reinforcement of the weak parts of the long-span transmission tower.

Modeling and modal analysis of the structure of long-span transmission tower

In a computer system the typically high overhead requirement for CPU instructions to operate a conventional direct memory (DMA) controler (2) are reduced to two user-level memory references via User-level Direct Memory Access (UDMA) (40). The UDMA apparatus (40) is located between the CPU and a DMA controller (2), whereby the UDMA is programmed to use existing virtual memory translation hardware of the associated computer system to perform permission checking and address translation without Kernel involvement, and otherwise use minimal Kernel involvement for other operations.

Method and system for implementing data transfers

The widespread commercialization of fuel cells relies on their improvement in safety and reliability. Suppressing the pressure difference between the two sides of the proton exchange membrane under dynamic conditions is challenging in developing fuel cell systems for transportation applications. This paper studies the coupling characteristics between pressure and mass flow in an air supply system. The results provide a basis for determining a control strategy to reduce pressure fluctuation on the membrane. Firstly, the open-loop frequency of the air supply system was tested through the experiments. Based on the results, an appropriate M-sequence was designed as an excitation signal of the air supply system, and the response output was obtained. Then, the least square method was used to identify the system transfer function for the reference operating point, and the pressure-mass flow coupling characteristics at different operating points were analyzed. Finally, a PI control with feedforward compensation decoupling was designed to achieve high-precision pressure and mass flow regulation.

Kai Li

Papers

PruMUX: Augmenting Data Multiplexing with Model Compression

Tail multi-parameter optimization of Ahmed model based on response surface method

Modeling and modal analysis of the structure of long-span transmission tower

Method and system for implementing data transfers

Research on Coupling Characteristics of Fuel Cell Air Supply System with Centrifugal Air Compressor