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 the commonly used representation of a feature-rich data object has evolved from a single feature vector to a set of feature vectors, a key challenge in building a content-based search engine for feature-rich data is to match feature-sets efficiently. Although substantial progress has been made during the past few years, existing approaches are still inefficient and inflexible for building a search engine for massive datasets. This paper presents a randomized algorithm to embed a set of features into a single high-dimensional vector to simplify the feature-set matching problem. The main idea is to project feature vectors into an auxiliary space using locality sensitive hashing and to represent a set of features as a histogram in the auxiliary space. A histogram is simply a high dimensional vector, and efficient similarity measures like L1 and L2 distances can be employed to approximate feature-set distance measures.We evaluated the proposed approach under three different task settings, i.e. content-based image search, image object recognition and near-duplicate video clip detection. The experimental results show that the proposed approach is indeed effective and flexible. It can achieve accuracy comparable to the feature-set matching methods, while requiring significantly less space and time. For object recognition with Caltech 101 dataset, our method runs 25 times faster to achieve the same precision as Pyramid Matching Kernel, the state-of-the-art feature-set matching method.

/pdf/efficiently-matching-sets-of-features-with-random-histograms-15hv0sixou.pdf

Efficiently matching sets of features with random histograms

The SHRIMP multicomputer provides virtual memory-mapped communication (VMMC), which supports protected, user-level message passing, allows user programs to perform their own buffer management, and separates data transfers from control transfers so that a data transfer can be done without the intervention of the receiving node CPU. An important question is whether such a mechanism can indeed deliver all of the available hardware performance to applications which use conventional message-passing libraries.This paper reports our early experience with message-passing on a small, working SHRIMP multicomputer. We have implemented several user-level communication libraries on top of the VMMC mechanism, including the NX message-passing interface, Sun RPC, stream sockets, and specialized RPC. The first three are fully compatible with existing systems. Our experience shows that the VMMC mechanism supports these message-passing interfaces well. When zero-copy protocols are allowed by the semantics of the interface, VMMC can effectively deliver to applications almost all of the raw hardware's communication performance.

https://dl.acm.org/doi/pdf/10.1145/232974.233004

Early Experience with Message-Passing on the SHRIMP Multicomputer

This paper examines how VI-based interconnects can be used to improve I/O path performance between a database server and the storage subsystem. We design and implement a software layer, DSA, that is layered between the application and VI. DSA takes advantage of specific VI features and deals with many of its shortcomings. We provide and evaluate one kernel-level and two user-level implementations of DSA. These implementations trade transparency and generality for performance at different degrees, and unlike research prototypes are designed to be suitable for real-world deployment. We present detailed measurements using a commercial database management system with both micro-benchmarks and industrial database workloads on a mid-size, 4 CPU, and a large, 32 CPU, database server.Our results show that VI-based interconnects and user-level communication can improve all aspects of the I/O path between the database system and the storage back-end. We also find that to make effective use of VI in I/O intensive environments we need to provide substantial additional functionality than what is currently provided by VI. Finally, new storage APIs that help minimize kernel involvement in the I/O path are needed to fully exploit the benefits of VI-based communication.

/pdf/experiences-with-vi-communication-for-database-storage-3r05nn2p90.pdf

Experiences with VI communication for database storage

In this paper we describe an experiment designed to evaluate the potential transaction processing system performance achievable through the combination of multiple processors and massive memories. The experiment consisted of the design and implementation of a transaction processing kernel on stock multiprocessors. We found that with sufficient memory, multiple processors can greatly improve performance. A prototype implementation of the kernel on a pair of Firefly multiprocessors (each with five 1-MIP processors) runs the standard debit-credit benchmark at over 1000 transactions per second.

Multiprocessor main memory transaction processing

This paper proposes a novel way to use virtual memorymapped communication (VMMC) to reduce the failover time on clusters. With the VMMC model, applications’ virtual address space can be efficiently mirrored on remote memory either automatically or via explicit messages. When a machine fails, its applications can restart from the most recent checkpoints on the failover node with minimal memory copying and disk I/O overhead. This method requires little change to applications’ source code. We developed two fast failover protocols: deliberate update failover protocol (DU) and automatic update failover protoco2 (AU). The first can run on any system that supports VMMC, whereas the other requires special network interface support. We implemented these two protocols on two different clusters that supported VMMC communication. Our results with three transaction-based applications show that both protocols work quite well. The deliberate update protocol imposes 4-21% overhead when taking checkpoints every 2 seconds. If an application can tolerate 20% overhead, this protocol can failover to another machine within 4 milliseconds in the best case and from 0.1 to 3 seconds in the worst case. The failover performance can be further improved by using special network interface hardware. The automatic update protocol is able to take checkpoints every 0.1 seconds with only 3-12% overhead. If 10% overhead is allowed, it can failover applications from 0.01 to 0.4 seconds in the worst case.

Kai Li

Papers

Efficiently matching sets of features with random histograms

Early Experience with Message-Passing on the SHRIMP Multicomputer

Experiences with VI communication for database storage

Multiprocessor main memory transaction processing

Fast cluster failover using virtual memory-mapped communication