Many statistical learning problems in the area of natural language processing including sequence tagging, sequence segmentation and syntactic parsing has been successfully approached by means of structured prediction methods. An appealing property of the corresponding discriminative learning algorithms is their ability to integrate the loss function of interest directly into the optimization process, which potentially can increase the resulting performance accuracy. Here, we demonstrate on the example of constituency parsing how to optimize for F1-score in the max-margin framework of structural SVM. In particular, the optimization is with respect to the original (not binarized) trees.
Structured prediction is ubiquitous in applications of machine learning such as knowledge extraction and natural language processing. Structure often can be formulated in terms of logical constraints. We consider the question of how to perform efficient active learning in the presence of logical constraints among variables inferred by different classifiers. We propose several methods and provide theoretical results that demonstrate the inappropriateness of employing uncertainty guided sampling, a commonly used active learning method. Furthermore, experiments on ten different datasets demonstrate that the methods significantly outperform alternatives in practice. The results are of practical significance in situations where labeled data is scarce.
Reward augmented maximum likelihood (RAML), a simple and effective learning framework to directly optimize towards the reward function in structured prediction tasks, has led to a number of impressive empirical successes. RAML incorporates task-specific reward by performing maximum-likelihood updates on candidate outputs sampled according to an exponentiated payoff distribution, which gives higher probabilities to candidates that are close to the reference output. While RAML is notable for its simplicity, efficiency, and its impressive empirical successes, the theoretical properties of RAML, especially the behavior of the exponentiated payoff distribution, has not been examined thoroughly. In this work, we introduce softmax Q-distribution estimation, a novel theoretical interpretation of RAML, which reveals the relation between RAML and Bayesian decision theory. The softmax Q-distribution can be regarded as a smooth approximation of the Bayes decision boundary, and the Bayes decision rule is achieved by decoding with this Q-distribution. We further show that RAML is equivalent to approximately estimating the softmax Q-distribution, with the temperature $\tau$ controlling approximation error. We perform two experiments, one on synthetic data of multi-class classification and one on real data of image captioning, to demonstrate the relationship between RAML and the proposed softmax Q-distribution estimation method, verifying our theoretical analysis. Additional experiments on three structured prediction tasks with rewards defined on sequential (named entity recognition), tree-based (dependency parsing) and irregular (machine translation) structures show notable improvements over maximum likelihood baselines.
We present a new technique for learning visual-semantic embeddings for cross-modal retrieval. Inspired by hard negative mining, the use of hard negatives in structured prediction, and ranking loss functions, we introduce a simple change to common loss functions used for multi-modal embeddings. That, combined with fine-tuning and use of augmented data, yields significant gains in retrieval performance. We showcase our approach, VSE++, on MS-COCO and Flickr30K datasets, using ablation studies and comparisons with existing methods. On MS-COCO our approach outperforms state-of-the-art methods by 8.8% in caption retrieval and 11.3% in image retrieval (at R@1).
Few-shot learning refers to understanding new concepts from only a few examples. We propose an information retrieval-inspired approach for this problem that is motivated by the increased importance of maximally leveraging all the available information in this low-data regime. We define a training objective that aims to extract as much information as possible from each training batch by effectively optimizing over all relative orderings of the batch points simultaneously. In particular, we view each batch point as a `query' that ranks the remaining ones based on its predicted relevance to them and we define a model within the framework of structured prediction to optimize mean Average Precision over these rankings. Our method achieves impressive results on the standard few-shot classification benchmarks while is also capable of few-shot retrieval.
Dialogue Act Recognition (DAR) is a challenging problem in dialogue interpretation, which aims to attach semantic labels to utterances and characterize the speaker's intention. Currently, many existing approaches formulate the DAR problem ranging from multi-classification to structured prediction, which suffer from handcrafted feature extensions and attentive contextual structural dependencies. In this paper, we consider the problem of DAR from the viewpoint of extending richer Conditional Random Field (CRF) structural dependencies without abandoning end-to-end training. We incorporate hierarchical semantic inference with memory mechanism on the utterance modeling. We then extend structured attention network to the linear-chain conditional random field layer which takes into account both contextual utterances and corresponding dialogue acts. The extensive experiments on two major benchmark datasets Switchboard Dialogue Act (SWDA) and Meeting Recorder Dialogue Act (MRDA) datasets show that our method achieves better performance than other state-of-the-art solutions to the problem. It is a remarkable fact that our method is nearly close to the human annotator's performance on SWDA within 2% gap.
We describe a technique for structured prediction, based on canonical correlation analysis. Our learning algorithm finds two projections for the input and the output spaces that aim at projecting a given input and its correct output into points close to each other. We demonstrate our technique on a language-vision problem, namely the problem of giving a textual description to an "abstract scene".
Parametric adversarial divergences, which are a generalization of the losses used to train generative adversarial networks (GANs), have often been described as being approximations of their nonparametric counterparts, such as the Jensen-Shannon divergence, which can be derived under the so-called optimal discriminator assumption. In this position paper, we argue that despite being "non-optimal", parametric divergences have distinct properties from their nonparametric counterparts which can make them more suitable for learning high-dimensional distributions. A key property is that parametric divergences are only sensitive to certain aspects/moments of the distribution, which depend on the architecture of the discriminator and the loss it was trained with. In contrast, nonparametric divergences such as the Kullback-Leibler divergence are sensitive to moments ignored by the discriminator, but they do not necessarily correlate with sample quality (Theis et al., 2016). Similarly, we show that mutual information can lead to unintuitive interpretations, and explore more intuitive alternatives based on parametric divergences. We conclude that parametric divergences are a flexible framework for defining statistical quantities relevant to a specific modeling task.
We introduce Graph-Structured Sum-Product Networks (GraphSPNs), a probabilistic approach to structured prediction for problems where dependencies between latent variables are expressed in terms of arbitrary, dynamic graphs. While many approaches to structured prediction place strict constraints on the interactions between inferred variables, many real-world problems can be only characterized using complex graph structures of varying size, often contaminated with noise when obtained from real data. Here, we focus on one such problem in the domain of robotics. We demonstrate how GraphSPNs can be used to bolster inference about semantic, conceptual place descriptions using noisy topological relations discovered by a robot exploring large-scale office spaces. Through experiments, we show that GraphSPNs consistently outperform the traditional approach based on undirected graphical models, successfully disambiguating information in global semantic maps built from uncertain, noisy local evidence. We further exploit the probabilistic nature of the model to infer marginal distributions over semantic descriptions of as yet unexplored places and detect spatial environment configurations that are novel and incongruent with the known evidence.
Recent systems on structured prediction focus on increasing the level of structural dependencies within the model. However, our study suggests that complex structures entail high overfitting risks. To control the structure-based overfitting, we propose to conduct structure regularization decoding (SR decoding). The decoding of the complex structure model is regularized by the additionally trained simple structure model. We theoretically analyze the quantitative relations between the structural complexity and the overfitting risk. The analysis shows that complex structure models are prone to the structure-based overfitting. Empirical evaluations show that the proposed method improves the performance of the complex structure models by reducing the structure-based overfitting. On the sequence labeling tasks, the proposed method substantially improves the performance of the complex neural network models. The maximum F1 error rate reduction is 36.4% for the third-order model. The proposed method also works for the parsing task. The maximum UAS improvement is 5.5% for the tri-sibling model. The results are competitive with or better than the state-of-the-art results.
Sequential decision making problems, such as structured prediction, robotic control, and game playing, require a combination of planning policies and generalisation of those plans. In this paper, we present Expert Iteration (ExIt), a novel reinforcement learning algorithm which decomposes the problem into separate planning and generalisation tasks. Planning new policies is performed by tree search, while a deep neural network generalises those plans. Subsequently, tree search is improved by using the neural network policy to guide search, increasing the strength of new plans. In contrast, standard deep Reinforcement Learning algorithms rely on a neural network not only to generalise plans, but to discover them too. We show that ExIt outperforms REINFORCE for training a neural network to play the board game Hex, and our final tree search agent, trained tabula rasa, defeats MoHex 1.0, the most recent Olympiad Champion player to be publicly released.
Many predicted structured objects (e.g., sequences, matchings, trees) are evaluated using the F-score, alignment error rate (AER), or other multivariate performance measures. Since inductively optimizing these measures using training data is typically computationally difficult, empirical risk minimization of surrogate losses is employed, using, e.g., the hinge loss for (structured) support vector machines. These approximations often introduce a mismatch between the learner's objective and the desired application performance, leading to inconsistency. We take a different approach: adversarially approximate training data while optimizing the exact F-score or AER. Structured predictions under this formulation result from solving zero-sum games between a predictor seeking the best performance and an adversary seeking the worst while required to (approximately) match certain structured properties of the training data. We explore this approach for word alignment (AER evaluation) and named entity recognition (F-score evaluation) with linear-chain constraints.
Are we using the right potential functions in the Conditional Random Field models that are popular in the Vision community? Semantic segmentation and other pixel-level labelling tasks have made significant progress recently due to the deep learning paradigm. However, most state-of-the-art structured prediction methods also include a random field model with a hand-crafted Gaussian potential to model spatial priors, label consistencies and feature-based image conditioning. In this paper, we challenge this view by developing a new inference and learning framework which can learn pairwise CRF potentials restricted only by their dependence on the image pixel values and the size of the support. Both standard spatial and high-dimensional bilateral kernels are considered. Our framework is based on the observation that CRF inference can be achieved via projected gradient descent and consequently, can easily be integrated in deep neural networks to allow for end-to-end training. It is empirically demonstrated that such learned potentials can improve segmentation accuracy and that certain label class interactions are indeed better modelled by a non-Gaussian potential. In addition, we compare our inference method to the commonly used mean-field algorithm. Our framework is evaluated on several public benchmarks for semantic segmentation with improved performance compared to previous state-of-the-art CNN+CRF models.
Deep learning methods have surpassed the performance of traditional techniques on a wide range of problems in computer vision, but nearly all of this work has studied consumer photos, where precisely correct output is often not critical. It is less clear how well these techniques may apply on structured prediction problems where fine-grained output with high precision is required, such as in scientific imaging domains. Here we consider the problem of segmenting echogram radar data collected from the polar ice sheets, which is challenging because segmentation boundaries are often very weak and there is a high degree of noise. We propose a multi-task spatiotemporal neural network that combines 3D ConvNets and Recurrent Neural Networks (RNNs) to estimate ice surface boundaries from sequences of tomographic radar images. We show that our model outperforms the state-of-the-art on this problem by (1) avoiding the need for hand-tuned parameters, (2) extracting multiple surfaces (ice-air and ice-bed) simultaneously, (3) requiring less non-visual metadata, and (4) being about 6 times faster.
In this paper, we propose a structured image inpainting method employing an energy based model. In order to learn structural relationship between patterns observed in images and missing regions of the images, we employ an energy-based structured prediction method. The structural relationship is learned by minimizing an energy function which is defined by a simple convolutional neural network. The experimental results on various benchmark datasets show that our proposed method significantly outperforms the state-of-the-art methods which use Generative Adversarial Networks (GANs). We obtained 497.35 mean squared error (MSE) on the Olivetti face dataset compared to 833.0 MSE provided by the state-of-the-art method. Moreover, we obtained 28.4 dB peak signal to noise ratio (PSNR) on the SVHN dataset and 23.53 dB on the CelebA dataset, compared to 22.3 dB and 21.3 dB, provided by the state-of-the-art methods, respectively. The code is publicly available.
We provide novel theoretical insights on structured prediction in the context of efficient convex surrogate loss minimization with consistency guarantees. For any task loss, we construct a convex surrogate that can be optimized via stochastic gradient descent and we prove tight bounds on the so-called "calibration function" relating the excess surrogate risk to the actual risk. In contrast to prior related work, we carefully monitor the effect of the exponential number of classes in the learning guarantees as well as on the optimization complexity. As an interesting consequence, we formalize the intuition that some task losses make learning harder than others, and that the classical 0-1 loss is ill-suited for general structured prediction.
Automated pavement crack detection is a challenging task that has been researched for decades due to the complicated pavement conditions in real world. In this paper, a supervised method based on deep learning is proposed, which has the capability of dealing with different pavement conditions. Specifically, a convolutional neural network (CNN) is used to learn the structure of the cracks from raw images, without any preprocessing. Small patches are extracted from crack images as inputs to generate a large training database, a CNN is trained and crack detection is modeled as a multi-label classification problem. Typically, crack pixels are much fewer than non-crack pixels. To deal with the problem with severely imbalanced data, a strategy with modifying the ratio of positive to negative samples is proposed. The method is tested on two public databases and compared with five existing methods. Experimental results show that it outperforms the other methods.
Structured prediction requires searching over a combinatorial number of structures. To tackle it, we introduce SparseMAP: a new method for sparse structured inference, and its natural loss function. SparseMAP automatically selects only a few global structures: it is situated between MAP inference, which picks a single structure, and marginal inference, which assigns probability mass to all structures, including implausible ones. Importantly, SparseMAP can be computed using only calls to a MAP oracle, making it applicable to problems with intractable marginal inference, e.g., linear assignment. Sparsity makes gradient backpropagation efficient regardless of the structure, enabling us to augment deep neural networks with generic and sparse structured hidden layers. Experiments in dependency parsing and natural language inference reveal competitive accuracy, improved interpretability, and the ability to capture natural language ambiguities, which is attractive for pipeline systems.
Many machine learning problems require the prediction of multi-dimensional labels. Such structured prediction models can benefit from modeling dependencies between labels. Recently, several deep learning approaches to structured prediction have been proposed. Here we focus on capturing cardinality constraints in such models. Namely, constraining the number of non-zero labels that the model outputs. Such constraints have proven very useful in previous structured prediction approaches, but it is a challenge to introduce them into a deep learning framework. Here we show how to do this via a novel deep architecture. Our approach outperforms strong baselines, achieving state-of-the-art results on multi-label classification benchmarks.
Dynamic programming (DP) solves a variety of structured combinatorial problems by iteratively breaking them down into smaller subproblems. In spite of their versatility, DP algorithms are usually non-differentiable, which hampers their use as a layer in neural networks trained by backpropagation. To address this issue, we propose to smooth the max operator in the dynamic programming recursion, using a strongly convex regularizer. This allows to relax both the optimal value and solution of the original combinatorial problem, and turns a broad class of DP algorithms into differentiable operators. Theoretically, we provide a new probabilistic perspective on backpropagating through these DP operators, and relate them to inference in graphical models. We derive two particular instantiations of our framework, a smoothed Viterbi algorithm for sequence prediction and a smoothed DTW algorithm for time-series alignment. We showcase these instantiations on two structured prediction tasks and on structured and sparse attention for neural machine translation.