Abstract

We consider the problem of weakly supervised object detection, where the training samples are annotated using only image-level labels that indicate the presence or absence of an object category. In order to model the uncertainty in the location of the objects, we employ a dissimilarity coefficient based probabilistic learning objective. The learning objective minimizes the difference between an annotation agnostic prediction distribution and an annotation aware conditional distribution. The main computational challenge is the complex nature of the conditional distribution, which consists of terms over hundreds or thousands of variables. The complexity of the conditional distribution rules out the possibility of explicitly modeling it. Instead, we exploit the fact that deep learning frameworks rely on stochastic optimization. This allows us to use a state of the art discrete generative model that can provide annotation consistent samples from the conditional distribution. Extensive experiments on PASCAL VOC 2007 and 2012 data sets demonstrate the efficacy of our proposed approach.

Abstract

Discrete labeling problems are often solvedby formulating them as an integer program,and relaxing the integrality constraint to acontinuous domain. While the continuous re-laxation is closely related to the original in-teger program, its optimal solution is oftenfractional. Thus, the success of a relaxationdepends crucially on the availability of an ac-curate rounding procedure. The problem ofidentifying an accurate rounding procedurehas mainly been tackled in the theoreticalcomputer science community through math-ematical analysis of the worst-case. However,this approach is both onerous and ignores thedistribution of the data encountered in prac-tice. We present a novel interpretation ofrounding procedures as sampling from a la-tent variable model, which opens the door tothe use of powerful machine learning formu-lations in their design. Inspired by the recentsuccess of deep latent variable models we pa-rameterize rounding procedures as a neuralnetwork, which lends itself to efficient opti-mization via back-propagation. By minimiz-ing the expected value of the objective of thediscrete labeling problem over training sam-ples, we learn a rounding procedure that ismore suited to the task at hand. Using bothsynthetic and real world data sets, we demon-strate that our approach can outperform thestate-of-the-art hand-designed rounding pro-cedures

Abstract

The accuracy of information retrieval systems is often measured using complex loss functions such as the average precision (AP) or the normalized discounted cumulative gain (NDCG). Given a set of positive and negative samples, the parameters of a retrieval system can be estimated by minimizing these loss functions. However, the non-differentiability and non-decomposability of these loss functions does not allow for simple gradient based optimization algorithms. This issue is generally circumvented by either optimizing a structured hinge-loss upper bound to the loss function or by using asymptotic methods like the direct-loss minimization framework. Yet, the high computational complexity of loss-augmented inference, which is necessary for both the frameworks, prohibits its use in large training data sets. To alleviate this deficiency, we present a novel quicksort flavored algorithm for a large class of non-decomposable loss functions. We provide a complete characterization of the loss functions that are amenable to our algorithm, and show that it includes both AP and NDCG based loss functions. Furthermore, we prove that no comparison based algorithm can improve upon the computational complexity of our approach asymptotically. We demonstrate the effectiveness of our approach in the context of optimizing the structured hinge loss upper bound of AP and NDCG loss for learning models for a variety of vision tasks. We show that our approach provides significantly better results than simpler decomposable loss functions, while requiring a comparable training time.

Abstract

We consider the task of learning to estimate human pose in still images. In order to avoid the high cost of full supervision, we propose to use a diverse data set, which consists of two types of annotations: (i) a small number of images are labeled using the expensive ground-truth pose; and (ii) other images are labeled using the inexpensive action label. As action information helps narrow down the pose of a human, we argue that this approach can help reduce the cost of training without significantly affecting the accuracy. To demonstrate this we design a probabilistic framework that employs two distributions: (i) a conditional distribution to model the uncertainty over the human pose given the image and the action; and (ii) a prediction distribution, which provides the pose of an image without using any action information. We jointly estimate the parameters of the two aforementioned distributions by minimizing their dissimilarity coefficient, as measured by a task-specific loss function. During both training and testing, we only require an efficient sampling strategy for both the aforementioned distributions. This allows us to use deep probabilistic networks that are capable of providing accurate pose estimates for previously unseen images. Using the MPII data set, we show that our approach outperforms baseline methods that either do not use the diverse annotations or rely on pointwise estimates of the pose.

Abstract

Sampark is machine translation system providing translations among nine pairs of Indian languages. Machine translation system is a class of natural language processing applications that is far more complex and highly compute intensive in nature. As the load on the deployed system increases, optimization becomes a challenge. Caching is one of the available options to improve the performance of a software system with increasing load which exhibit the characteristics of locality of reference. Sampark MT system being a natural language processing application exhibits this characteristic.Memcachedis a well-known, simple, in-memory caching solution that has been applied to improve the performance of several distributed web applications in the past. This paper describes how memcached has been applied to improve the performance of Sampark machine translation service which is deployed on a large cluster of machines. By applying distributed caching to MT system the performance of the system has improved upto 40%

Abstract

We propose a novel partial linearization based approach for optimizing the multi-class svm learning problem. Our method is an intuitive generalization of the Frank-Wolfe and the exponentiated gradient algorithms. In particular, it allows us to combine several of their desirable qualities into one approach: (i) the use of an expectation oracle (which provides the marginals over each output class) in order to estimate an informative descent direction, similar to exponentiated gradient; (ii) analytical computation of the optimal step-size in the descent direction that guarantees an increase in the dual objective, similar to Frank-Wolfe; and (iii) a block coordinate formulation similar to the one proposed for Frank-Wolfe, which allows us to solve large-scale problems. Using the challenging computer vision problems of action classification, object recognition and gesture recognition, we demonstrate the efficacy of our …

Abstract

The performance of binary classification tasks, such as action classification and object detection, is often measured in terms of the average precision (AP). Yet it is common practice in computer vision to employ the support vector machine (SVM) classifier, which optimizes a surrogate 0-1 loss. The popularity of SVM can be attributed to its empirical performance. Specifically, in fully supervised settings, SVM tends to provide similar accuracy to the AP-SVM classifier, which directly optimizes an AP-based loss. However, we hypothesize that in the significantly more challenging and practically useful setting of weakly supervised learning, it becomes crucial to optimize the right accuracy measure. In order to test this hypothesis, we propose a novel latent AP-SVM that minimizes a carefully designed upper bound on the AP-based loss function over weakly supervised samples. Using publicly available datasets, we demonstrate the advantage of our approach over standard loss-based binary classifiers on two challenging problems: action classification and character recognition.

Abstract

The accuracy of information retrieval systems is often measured using average precision (AP). Given a set of positive (relevant) and negative (non-relevant) samples, the parameters of a retrieval system can be estimated using the AP-SVM framework, which minimizes a regularized convex upper bound on the empirical AP loss. However, the high computational complexity of loss-augmented inference, which is required for learning an AP-SVM, prohibits its use with large training datasets. To alleviate this deficiency, we propose three complementary approaches. The first approach guarantees an asymptotic decrease in the computational complexity of loss-augmented inference by exploiting the problem structure. The second approach takes advantage of the fact that we do not require a full ranking during loss-augmented inference. This helps us to avoid the expensive step of sorting the negative samples according to their individual scores. The third approach approximates the AP loss over all samples by the AP loss over difficult samples (for example, those that are incorrectly classified by a binary SVM), while ensuring the correct classification of the remaining samples. Using the PASCAL VOC action classification and object detection datasets, we show that our approaches provide significant speed-ups during training without degrading the test accuracy of AP-SVM.

Abstract

The problem of ranking a set of visual samples according to their relevance to a query plays an important role in computer vision. The traditional approach for ranking is to train a binary classifier such as a support vector machine (svm). Binary classifiers suffer from two main deficiencies: (i) they do not optimize a ranking-based loss function, for example, the average precision (ap) loss; and (ii) they cannot incorporate high-order information such as the a priori correlation between the relevance of two visual samples (for example, two persons in the same image tend to perform the same action). We propose two novel learning formulations that allow us to incorporate high-order information for ranking. The first framework, called high-order binary svm (hob-svm), allows for a structured input. The parameters of hob-svm are learned by minimizing a convex upper bound on a surrogate 0-1 loss function. In order to obtain the ranking of the samples that form the structured input, hobsvm sorts the samples according to their max-marginals. The second framework, called high-order average precision svm (hoap-svm), also allows for a structured input and uses the same ranking criterion. However, in contrast to hob-svm, the parameters of hoap-svm are learned by minimizing a difference-of-convex upper bound on the ap loss. Using a standard, publicly available dataset for the challenging problem of action classification, we show that both hob-svm and hoap-svm outperform the baselines that ignore high-order information.

Abstract

Displays remain flat and passive amidst the many changes in their fundamental technologies. One natural step ahead is tocreate displays that merge seamlessly in shape and appearance with one’s natural surroundings. In this paper, we present a system todesign, render to, and build view-dependentmultiplanar displaysof arbitrary piecewise-planar shapes, built using polygonal facets.Our system provides high quality, interactive rendering of 3D environments to a head-tracked viewer on arbitrary multiplanar displays.We develop a novel rendering scheme that produces exact image and depth map at each facet, producing artifact-free images on andacross facet boundaries. The system scales to a large number of display facets by rendering all facets in a single pass of rasterization.This is achieved using a parallel, perframe, view-dependent binning and prewarping of scene triangles. The display is driven using oneor more targetquilt imagesinto which facet pixels are packed. Our method places no constraints on the scene or the display and allowsfor fully dynamic scenes to be rendered interactively at high resolutions. The steps of our system are implemented efficiently oncommodity GPUs. We present a few prototype displays to establish the scalability of our system on different display shapes, formfactors, and complexity: from a cube made out of LCD panels to spherical/cylindrical projected setups to arbitrary complex shapes insimulation. Performance of our system is demonstrated for both rendering quality and speed, for increasing scene and display facetsizes. A subjective user study is also presented to evaluate the user experience using a walk-around display compared to a flat panelfor a game-like setting.