scispace - formally typeset
Search or ask a question
Proceedings ArticleDOI

No Fuss Distance Metric Learning Using Proxies

Yair Movshovitz-Attias1, Alexander Toshev1, Thomas Leung1, Sergey Ioffe1, Saurabh Singh1 
01 Oct 2017-pp 360-368
TL;DR: This paper proposes to optimize the triplet loss on a different space of triplets, consisting of an anchor data point and similar and dissimilar proxy points which are learned as well, and proposes a proxy-based loss which improves on state-of-art results for three standard zero-shot learning datasets.
Abstract: We address the problem of distance metric learning (DML), defined as learning a distance consistent with a notion of semantic similarity. Traditionally, for this problem supervision is expressed in the form of sets of points that follow an ordinal relationship – an anchor point x is similar to a set of positive points Y , and dissimilar to a set of negative points Z, and a loss defined over these distances is minimized. While the specifics of the optimization differ, in this work we collectively call this type of supervision Triplets and all methods that follow this pattern Triplet-Based methods. These methods are challenging to optimize. A main issue is the need for finding informative triplets, which is usually achieved by a variety of tricks such as increasing the batch size, hard or semi-hard triplet mining, etc. Even with these tricks, the convergence rate of such methods is slow. In this paper we propose to optimize the triplet loss on a different space of triplets, consisting of an anchor data point and similar and dissimilar proxy points which are learned as well. These proxies approximate the original data points, so that a triplet loss over the proxies is a tight upper bound of the original loss. This proxy-based loss is empirically better behaved. As a result, the proxy-loss improves on state-of-art results for three standard zero-shot learning datasets, by up to 15% points, while converging three times as fast as other triplet-based losses.
Citations
More filters
Proceedings ArticleDOI
15 Jun 2019
TL;DR: In this article, a general pair weighting (GPW) framework has been proposed, which casts the sampling problem of deep metric learning into a unified view through gradient analysis, providing a powerful tool for understanding recent pair-based loss functions.
Abstract: A family of loss functions built on pair-based computation have been proposed in the literature which provide a myriad of solutions for deep metric learning. In this pa-per, we provide a general weighting framework for under-standing recent pair-based loss functions. Our contributions are three-fold: (1) we establish a General Pair Weighting (GPW) framework, which casts the sampling problem of deep metric learning into a unified view of pair weighting through gradient analysis, providing a powerful tool for understanding recent pair-based loss functions; (2) we show that with GPW, various existing pair-based methods can be compared and discussed comprehensively, with clear differences and key limitations identified; (3) we propose a new loss called multi-similarity loss (MS loss) under the GPW,which is implemented in two iterative steps (i.e., mining and weighting). This allows it to fully consider three similarities for pair weighting, providing a more principled approach for collecting and weighting informative pairs. Finally, the proposed MS loss obtains new state-of-the-art performance on four image retrieval benchmarks, where it outperforms the most recent approaches, such as ABE[14] and HTL[4], by a large margin, e.g.,60.6%→65.7%on CUB200,and 80.9%→88.0%on In-Shop Clothes Retrieval datasetat Recall@1.

549 citations

Proceedings ArticleDOI
18 Jun 2018
TL;DR: The process weight imprinting is called as it directly sets weights for a new category based on an appropriately scaled copy of the embedding layer activations for that training example, which provides immediate good classification performance and an initialization for any further fine-tuning in the future.
Abstract: Human vision is able to immediately recognize novel visual categories after seeing just one or a few training examples. We describe how to add a similar capability to ConvNet classifiers by directly setting the final layer weights from novel training examples during low-shot learning. We call this process weight imprinting as it directly sets weights for a new category based on an appropriately scaled copy of the embedding layer activations for that training example. The imprinting process provides a valuable complement to training with stochastic gradient descent, as it provides immediate good classification performance and an initialization for any further fine-tuning in the future. We show how this imprinting process is related to proxy-based embeddings. However, it differs in that only a single imprinted weight vector is learned for each novel category, rather than relying on a nearest-neighbor distance to training instances as typically used with embedding methods. Our experiments show that using averaging of imprinted weights provides better generalization than using nearest-neighbor instance embeddings.

540 citations


Cites background or methods from "No Fuss Distance Metric Learning Us..."

  • ...The implementation details are comparable to [11] and [12]....

    [...]

  • ...Unlike in [11] where only the embedding extractor part is used during test time with the auxiliary proxies thrown away, we keep the entirety of the network....

    [...]

  • ...[11] reformulated the loss by assigning proxies p(·) to training examples according to the class labels...

    [...]

  • ...Imprinted weights and Proxy-NCA are both trained with the softmax cross-entropy loss....

    [...]

  • ...The proxy-based loss, from which we have derived our method, upper bounds the instance-based triplet loss [11]....

    [...]

Proceedings ArticleDOI
15 Jun 2019
TL;DR: A novel instance based softmax embedding method, which directly optimizes the `real' instance features on top of the softmax function, which achieves significantly faster learning speed and higher accuracy than all existing methods.
Abstract: This paper studies the unsupervised embedding learning problem, which requires an effective similarity measurement between samples in low-dimensional embedding space. Motivated by the positive concentrated and negative separated properties observed from category-wise supervised learning, we propose to utilize the instance-wise supervision to approximate these properties, which aims at learning data augmentation invariant and instance spread-out features. To achieve this goal, we propose a novel instance based softmax embedding method, which directly optimizes the `real' instance features on top of the softmax function. It achieves significantly faster learning speed and higher accuracy than all existing methods. The proposed method performs well for both seen and unseen testing categories with cosine similarity. It also achieves competitive performance even without pre-trained network over samples from fine-grained categories.

341 citations


Cites background or methods from "No Fuss Distance Metric Learning Us..."

  • ...Supervised embedding learning methods have been studied to achieve such objectives and demonstrate impressive capabilities in various vision tasks [28, 30, 53]....

    [...]

  • ...Most of them are designed on top of pairwise [12, 30] or triplet relationships [13, 29]....

    [...]

  • ...The pre-trained Inception-V1 [39] on ImageNet is used as the backbone network following existing methods [30, 32, 37]....

    [...]

  • ...In testing phase, a single center-cropped image is adopted for fine-grained recognition as in [30]....

    [...]

  • ...For data augmentation, the images are randomly cropped at size 227×227 with random horizontal flipping following [21, 30]....

    [...]

Book ChapterDOI
23 Aug 2020
TL;DR: This work proposes PODNet, a model inspired by representation learning that fights catastrophic forgetting, even over very long runs of small incremental tasks --a setting so far unexplored by current works.
Abstract: Lifelong learning has attracted much attention, but existing works still struggle to fight catastrophic forgetting and accumulate knowledge over long stretches of incremental learning. In this work, we propose PODNet, a model inspired by representation learning. By carefully balancing the compromise between remembering the old classes and learning new ones, PODNet fights catastrophic forgetting, even over very long runs of small incremental tasks – a setting so far unexplored by current works. PODNet innovates on existing art with an efficient spatial-based distillation-loss applied throughout the model and a representation comprising multiple proxy vectors for each class. We validate those innovations thoroughly, comparing PODNet with three state-of-the-art models on three datasets: CIFAR100, ImageNet100, and ImageNet1000. Our results showcase a significant advantage of PODNet over existing art, with accuracy gains of 12.10, 6.51, and 2.85 percentage points, respectively.

306 citations


Cites background from "No Fuss Distance Metric Learning Us..."

  • ...A simple cross-entropy loss would work, but we found empirically that the NCA loss [10,28] converged faster....

    [...]

  • ...For each class, the normalized weight vector acts as a single proxy [28], towards which the learning procedure pushes all samples in the class....

    [...]

Proceedings ArticleDOI
01 Oct 2019
TL;DR: In this article, the authors proposed to directly optimize the global mAP by leveraging recent advances in listwise loss formulations, using a histogram binning approximation, which can be differentiated and thus employed to end-to-end learning.
Abstract: Image retrieval can be formulated as a ranking problem where the goal is to order database images by decreasing similarity to the query. Recent deep models for image retrieval have outperformed traditional methods by leveraging ranking-tailored loss functions, but important theoretical and practical problems remain. First, rather than directly optimizing the global ranking, they minimize an upper-bound on the essential loss, which does not necessarily result in an optimal mean average precision (mAP). Second, these methods require significant engineering efforts to work well, e.g., special pre-training and hard-negative mining. In this paper we propose instead to directly optimize the global mAP by leveraging recent advances in listwise loss formulations. Using a histogram binning approximation, the AP can be differentiated and thus employed to end-to-end learning. Compared to existing losses, the proposed method considers thousands of images simultaneously at each iteration and eliminates the need for ad hoc tricks. It also establishes a new state of the art on many standard retrieval benchmarks. Models and evaluation scripts have been made available at: https://europe.naverlabs.com/Deep-Image-Retrieval/.

301 citations

References
More filters
Proceedings ArticleDOI
07 Jun 2015
TL;DR: Inception as mentioned in this paper is a deep convolutional neural network architecture that achieves the new state of the art for classification and detection in the ImageNet Large-Scale Visual Recognition Challenge 2014 (ILSVRC14).
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.

40,257 citations

Journal ArticleDOI
TL;DR: The ImageNet Large Scale Visual Recognition Challenge (ILSVRC) as mentioned in this paper is a benchmark in object category classification and detection on hundreds of object categories and millions of images, which has been run annually from 2010 to present, attracting participation from more than fifty institutions.
Abstract: The ImageNet Large Scale Visual Recognition Challenge is a benchmark in object category classification and detection on hundreds of object categories and millions of images. The challenge has been run annually from 2010 to present, attracting participation from more than fifty institutions. This paper describes the creation of this benchmark dataset and the advances in object recognition that have been possible as a result. We discuss the challenges of collecting large-scale ground truth annotation, highlight key breakthroughs in categorical object recognition, provide a detailed analysis of the current state of the field of large-scale image classification and object detection, and compare the state-of-the-art computer vision accuracy with human accuracy. We conclude with lessons learned in the 5 years of the challenge, and propose future directions and improvements.

30,811 citations

Posted Content
Sergey Ioffe1, Christian Szegedy1
TL;DR: Batch Normalization as mentioned in this paper normalizes layer inputs for each training mini-batch to reduce the internal covariate shift in deep neural networks, and achieves state-of-the-art performance on ImageNet.
Abstract: Training Deep Neural Networks is complicated by the fact that the distribution of each layer's inputs changes during training, as the parameters of the previous layers change. This slows down the training by requiring lower learning rates and careful parameter initialization, and makes it notoriously hard to train models with saturating nonlinearities. We refer to this phenomenon as internal covariate shift, and address the problem by normalizing layer inputs. Our method draws its strength from making normalization a part of the model architecture and performing the normalization for each training mini-batch. Batch Normalization allows us to use much higher learning rates and be less careful about initialization. It also acts as a regularizer, in some cases eliminating the need for Dropout. Applied to a state-of-the-art image classification model, Batch Normalization achieves the same accuracy with 14 times fewer training steps, and beats the original model by a significant margin. Using an ensemble of batch-normalized networks, we improve upon the best published result on ImageNet classification: reaching 4.9% top-5 validation error (and 4.8% test error), exceeding the accuracy of human raters.

17,184 citations

Book
01 Jan 2008
TL;DR: In this article, the authors present an up-to-date treatment of all aspects of the design and implementation of systems for gathering, indexing, and searching documents; methods for evaluating systems; and an introduction to the use of machine learning methods on text collections.
Abstract: Class-tested and coherent, this groundbreaking new textbook teaches web-era information retrieval, including web search and the related areas of text classification and text clustering from basic concepts. Written from a computer science perspective by three leading experts in the field, it gives an up-to-date treatment of all aspects of the design and implementation of systems for gathering, indexing, and searching documents; methods for evaluating systems; and an introduction to the use of machine learning methods on text collections. All the important ideas are explained using examples and figures, making it perfect for introductory courses in information retrieval for advanced undergraduates and graduate students in computer science. Based on feedback from extensive classroom experience, the book has been carefully structured in order to make teaching more natural and effective. Although originally designed as the primary text for a graduate or advanced undergraduate course in information retrieval, the book will also create a buzz for researchers and professionals alike.

11,804 citations

Posted Content
TL;DR: The TensorFlow interface and an implementation of that interface that is built at Google are described, which has been used for conducting research and for deploying machine learning systems into production across more than a dozen areas of computer science and other fields.
Abstract: TensorFlow is an interface for expressing machine learning algorithms, and an implementation for executing such algorithms. A computation expressed using TensorFlow can be executed with little or no change on a wide variety of heterogeneous systems, ranging from mobile devices such as phones and tablets up to large-scale distributed systems of hundreds of machines and thousands of computational devices such as GPU cards. The system is flexible and can be used to express a wide variety of algorithms, including training and inference algorithms for deep neural network models, and it has been used for conducting research and for deploying machine learning systems into production across more than a dozen areas of computer science and other fields, including speech recognition, computer vision, robotics, information retrieval, natural language processing, geographic information extraction, and computational drug discovery. This paper describes the TensorFlow interface and an implementation of that interface that we have built at Google. The TensorFlow API and a reference implementation were released as an open-source package under the Apache 2.0 license in November, 2015 and are available at www.tensorflow.org.

10,447 citations


"No Fuss Distance Metric Learning Us..." refers methods in this paper

  • ...We used the TensorFlow Deep Learning framework [1] for all methods described below....

    [...]