The causal structure of any system can be analyzed at a multitude of spatial and temporal scales. It has long been thought that while higher scale (macro) descriptions of causal structure may be useful to observers, they are at best a compressed description and at worse leave out critical information. However, recent research applying information theory to causal analysis has shown that the causal structure of some systems can actually come into focus (be more informative) at a macroscale (Hoel et al. 2013). That is, a macro model of a system (a map) can be more informative than a fully detailed model of the system (the territory). This has been called causal emergence. While causal emergence may at first glance seem counterintuitive, this paper grounds the phenomenon in a classic concept from information theory: Shannon's discovery of the channel capacity. I argue that systems have a particular causal capacity, and that different causal models of those systems take advantage of that capacity to various degrees. For some systems, only macroscale causal models use the full causal capacity. Such macroscale causal models can either be coarse-grains, or may leave variables and states out of the model (exogenous) in various ways, which can improve the model's efficacy and its informativeness via the same mathematical principles of how error-correcting codes take advantage of an information channel's capacity. As model choice increase, the causal capacity of a system approaches the channel capacity. Ultimately, this provides a general framework for understanding how the causal structure of some systems cannot be fully captured by even the most detailed microscopic model.
Long-running, high-impact events such as the Boston Marathon bombing often develop through many stages and involve a large number of entities in their unfolding. Timeline summarization of an event by key sentences eases story digestion, but does not distinguish between what a user remembers and what she might want to re-check. In this work, we present a novel approach for timeline summarization of high-impact events, which uses entities instead of sentences for summarizing the event at each individual point in time. Such entity summaries can serve as both (1) important memory cues in a retrospective event consideration and (2) pointers for personalized event exploration. In order to automatically create such summaries, it is crucial to identify the "right" entities for inclusion. We propose to learn a ranking function for entities, with a dynamically adapted trade-off between the in-document salience of entities and the informativeness of entities across documents, i.e., the level of new information associated with an entity for a time point under consideration. Furthermore, for capturing collective attention for an entity we use an innovative soft labeling approach based on Wikipedia. Our experiments on a real large news datasets confirm the effectiveness of the proposed methods.
In recent years two sets of planar (2D) shape attributes, provided with an intuitive physical meaning, were proposed to the remote sensing community by, respectively, Nagao & Matsuyama and Shackelford & Davis in their seminal works on the increasingly popular geographic object based image analysis (GEOBIA) paradigm. These two published sets of intuitive geometric features were selected as initial conditions by the present R&D software project, whose multi-objective goal was to accomplish: (i) a minimally dependent and maximally informative design (knowledge/information representation) of a general purpose, user and application independent dictionary of 2D shape terms provided with a physical meaning intuitive to understand by human end users and (ii) an effective (accurate, scale invariant, easy to use) and efficient implementation of 2D shape descriptors. To comply with the Quality Assurance Framework for Earth Observation guidelines, the proposed suite of geometric functions is validated by means of a novel quantitative quality assurance policy, centered on inter feature dependence (causality) assessment. This innovative multivariate feature validation strategy is alternative to traditional feature selection procedures based on either inductive data learning classification accuracy estimation, which is inherently case specific, or cross correlation estimation, because statistical cross correlation does not imply causation. The project deliverable is an original general purpose software suite of seven validated off the shelf 2D shape descriptors intuitive to use. Alternative to existing commercial or open source software libraries of tens of planar shape functions whose informativeness remains unknown, it is eligible for use in (GE)OBIA systems in operating mode, expected to mimic human reasoning based on a convergence of evidence approach.
In this paper, we propose a novel approach for aggregating online reviews, according to the opinions they express. Our methodology is unsupervised - due to the fact that it does not rely on pre-labeled reviews - and it is agnostic - since it does not make any assumption about the domain or the language of the review content. We measure the adherence of a review content to the domain terminology extracted from a review set. First, we demonstrate the informativeness of the adherence metric with respect to the score associated with a review. Then, we exploit the metric values to group reviews, according to the opinions they express. Our experimental campaign has been carried out on two large datasets collected from Booking and Amazon, respectively.
Many real-world time-series analysis problems are characterised by scarce data. Solutions typically rely on hand-crafted features extracted from the time or frequency domain allied with classification or regression engines which condition on this (often low-dimensional) feature vector. The huge advances enjoyed by many application domains in recent years have been fuelled by the use of deep learning architectures trained on large data sets. This paper presents an application of deep learning for acoustic event detection in a challenging, data-scarce, real-world problem. Our candidate challenge is to accurately detect the presence of a mosquito from its acoustic signature. We develop convolutional neural networks (CNNs) operating on wavelet transformations of audio recordings. Furthermore, we interrogate the network's predictive power by visualising statistics of network-excitatory samples. These visualisations offer a deep insight into the relative informativeness of components in the detection problem. We include comparisons with conventional classifiers, conditioned on both hand-tuned and generic features, to stress the strength of automatic deep feature learning. Detection is achieved with performance metrics significantly surpassing those of existing algorithmic methods, as well as marginally exceeding those attained by individual human experts.
The learning of mixture models can be viewed as a clustering problem. Indeed, given data samples independently generated from a mixture of distributions, we often would like to find the {\it correct target clustering} of the samples according to which component distribution they were generated from. For a clustering problem, practitioners often choose to use the simple $k$-means algorithm. $k$-means attempts to find an {\it optimal clustering} that minimizes the sum-of-squares distance between each point and its cluster center. In this paper, we consider fundamental (i.e., information-theoretic) limits of the solutions (clusterings) obtained by optimizing the sum-of-squares distance. In particular, we provide sufficient conditions for the closeness of any optimal clustering and the correct target clustering assuming that the data samples are generated from a mixture of spherical Gaussian distributions. We also generalize our results to log-concave distributions. Moreover, we show that under similar or even weaker conditions on the mixture model, any optimal clustering for the samples with reduced dimensionality is also close to the correct target clustering. These results provide intuition for the informativeness of $k$-means (with and without dimensionality reduction) as an algorithm for learning mixture models.
Most extractive summarization methods focus on the main body of the document from which sentences need to be extracted. However, the gist of the document may lie in side information, such as the title and image captions which are often available for newswire articles. We propose to explore side information in the context of single-document extractive summarization. We develop a framework for single-document summarization composed of a hierarchical document encoder and an attention-based extractor with attention over side information. We evaluate our model on a large scale news dataset. We show that extractive summarization with side information consistently outperforms its counterpart that does not use any side information, in terms of both informativeness and fluency.
A growing number of applications, e.g. video surveillance and medical image analysis, require training recognition systems from large amounts of weakly annotated data while some targeted interactions with a domain expert are allowed to improve the training process. In such cases, active learning (AL) can reduce labeling costs for training a classifier by querying the expert to provide the labels of most informative instances. This paper focuses on AL methods for instance classification problems in multiple instance learning (MIL), where data is arranged into sets, called bags, that are weakly labeled. Most AL methods focus on single instance learning problems. These methods are not suitable for MIL problems because they cannot account for the bag structure of data. In this paper, new methods for bag-level aggregation of instance informativeness are proposed for multiple instance active learning (MIAL). The \textit{aggregated informativeness} method identifies the most informative instances based on classifier uncertainty, and queries bags incorporating the most information. The other proposed method, called \textit{cluster-based aggregative sampling}, clusters data hierarchically in the instance space. The informativeness of instances is assessed by considering bag labels, inferred instance labels, and the proportion of labels that remain to be discovered in clusters. Both proposed methods significantly outperform reference methods in extensive experiments using benchmark data from several application domains. Results indicate that using an appropriate strategy to address MIAL problems yields a significant reduction in the number of queries needed to achieve the same level of performance as single instance AL methods.
We introduce the Machine Translation (MT) evaluation survey that contains both manual and automatic evaluation methods. The traditional human evaluation criteria mainly include the intelligibility, fidelity, fluency, adequacy, comprehension, and informativeness. The advanced human assessments include task-oriented measures, post-editing, segment ranking, and extended criteriea, etc. We classify the automatic evaluation methods into two categories, including lexical similarity scenario and linguistic features application. The lexical similarity methods contain edit distance, precision, recall, F-measure, and word order. The linguistic features can be divided into syntactic features and semantic features respectively. The syntactic features include part of speech tag, phrase types and sentence structures, and the semantic features include named entity, synonyms, textual entailment, paraphrase, semantic roles, and language models. The deep learning models for evaluation are very newly proposed. Subsequently, we also introduce the evaluation methods for MT evaluation including different correlation scores, and the recent quality estimation (QE) tasks for MT. This paper differs from the existing works \cite{GALEprogram2009,EuroMatrixProject2007} from several aspects, by introducing some recent development of MT evaluation measures, the different classifications from manual to automatic evaluation measures, the introduction of recent QE tasks of MT, and the concise construction of the content. We hope this work will be helpful for MT researchers to easily pick up some metrics that are best suitable for their specific MT model development, and help MT evaluation researchers to get a general clue of how MT evaluation research developed. Furthermore, hopefully, this work can also shine some light on other evaluation tasks, except for translation, of NLP fields.
We present a novel training framework for neural sequence models, particularly for grounded dialog generation. The standard training paradigm for these models is maximum likelihood estimation (MLE), or minimizing the cross-entropy of the human responses. Across a variety of domains, a recurring problem with MLE trained generative neural dialog models (G) is that they tend to produce 'safe' and generic responses ("I don't know", "I can't tell"). In contrast, discriminative dialog models (D) that are trained to rank a list of candidate human responses outperform their generative counterparts; in terms of automatic metrics, diversity, and informativeness of the responses. However, D is not useful in practice since it cannot be deployed to have real conversations with users. Our work aims to achieve the best of both worlds -- the practical usefulness of G and the strong performance of D -- via knowledge transfer from D to G. Our primary contribution is an end-to-end trainable generative visual dialog model, where G receives gradients from D as a perceptual (not adversarial) loss of the sequence sampled from G. We leverage the recently proposed Gumbel-Softmax (GS) approximation to the discrete distribution -- specifically, an RNN augmented with a sequence of GS samplers, coupled with the straight-through gradient estimator to enable end-to-end differentiability. We also introduce a stronger encoder for visual dialog, and employ a self-attention mechanism for answer encoding along with a metric learning loss to aid D in better capturing semantic similarities in answer responses. Overall, our proposed model outperforms state-of-the-art on the VisDial dataset by a significant margin (2.67% on recall@10). The source code can be downloaded from https://github.com/jiasenlu/visDial.pytorch.
Unlike extractive summarization, abstractive summarization has to fuse different parts of the source text, which inclines to create fake facts. Our preliminary study reveals nearly 30% of the outputs from a state-of-the-art neural summarization system suffer from this problem. While previous abstractive summarization approaches usually focus on the improvement of informativeness, we argue that faithfulness is also a vital prerequisite for a practical abstractive summarization system. To avoid generating fake facts in a summary, we leverage open information extraction and dependency parse technologies to extract actual fact descriptions from the source text. The dual-attention sequence-to-sequence framework is then proposed to force the generation conditioned on both the source text and the extracted fact descriptions. Experiments on the Gigaword benchmark dataset demonstrate that our model can greatly reduce fake summaries by 80%. Notably, the fact descriptions also bring significant improvement on informativeness since they often condense the meaning of the source text.
A hallmark of human intelligence is the ability to ask rich, creative, and revealing questions. Here we introduce a cognitive model capable of constructing human-like questions. Our approach treats questions as formal programs that, when executed on the state of the world, output an answer. The model specifies a probability distribution over a complex, compositional space of programs, favoring concise programs that help the agent learn in the current context. We evaluate our approach by modeling the types of open-ended questions generated by humans who were attempting to learn about an ambiguous situation in a game. We find that our model predicts what questions people will ask, and can creatively produce novel questions that were not present in the training set. In addition, we compare a number of model variants, finding that both question informativeness and complexity are important for producing human-like questions.
Despite significant progress in a variety of vision-and-language problems, developing a method capable of asking intelligent, goal-oriented questions about images is proven to be an inscrutable challenge. Towards this end, we propose a Deep Reinforcement Learning framework based on three new intermediate rewards, namely goal-achieved, progressive and informativeness that encourage the generation of succinct questions, which in turn uncover valuable information towards the overall goal. By directly optimizing for questions that work quickly towards fulfilling the overall goal, we avoid the tendency of existing methods to generate long series of insane queries that add little value. We evaluate our model on the GuessWhat?! dataset and show that the resulting questions can help a standard Guesser identify a specific object in an image at a much higher success rate.
Active learning - a class of algorithms that iteratively searches for the most informative samples to include in a training dataset - has been shown to be effective at annotating data for image classification. However, the use of active learning for object detection is still largely unexplored as determining informativeness of an object-location hypothesis is more difficult. In this paper, we address this issue and present two metrics for measuring the informativeness of an object hypothesis, which allow us to leverage active learning to reduce the amount of annotated data needed to achieve a target object detection performance. Our first metric measures 'localization tightness' of an object hypothesis, which is based on the overlapping ratio between the region proposal and the final prediction. Our second metric measures 'localization stability' of an object hypothesis, which is based on the variation of predicted object locations when input images are corrupted by noise. Our experimental results show that by augmenting a conventional active-learning algorithm designed for classification with the proposed metrics, the amount of labeled training data required can be reduced up to 25%. Moreover, on PASCAL 2007 and 2012 datasets our localization-stability method has an average relative improvement of 96.5% and 81.9% over the baseline method using classification only.
Research on summarization has mainly been driven by empirical approaches, crafting systems to perform well on standard datasets with the notion of information Importance remaining latent. We argue that establishing theoretical models of Importance will advance our understanding of the task and help to further improve summarization systems. To this end, we propose simple but rigorous definitions of several concepts that were previously used only intuitively in summarization: Redundancy, Relevance, and Informativeness. Importance arises as a single quantity naturally unifying these concepts. Additionally, we provide intuitions to interpret the proposed quantities and experiments to demonstrate the potential of the framework to inform and guide subsequent works.
Crisis responders are increasingly using social media, data and other digital sources of information to build a situational understanding of a crisis situation in order to design an effective response. However with the increased availability of such data, the challenge of identifying relevant information from it also increases. This paper presents a successful automatic approach to handling this problem. Messages are filtered for informativeness based on a definition of the concept drawn from prior research and crisis response experts. Informative messages are tagged for actionable data -- for example, people in need, threats to rescue efforts, changes in environment, and so on. In all, eight categories of actionability are identified. The two components -- informativeness and actionability classification -- are packaged together as an openly-available tool called Emina (Emergent Informativeness and Actionability).
Active learning aims to train a classifier as fast as possible with as few labels as possible. The core element in virtually any active learning strategy is the criterion that measures the usefulness of the unlabeled data based on which new points to be labeled are picked. We propose a novel approach which we refer to as maximizing variance for active learning or MVAL for short. MVAL measures the value of unlabeled instances by evaluating the rate of change of output variables caused by changes in the next sample to be queried and its potential labelling. In a sense, this criterion measures how unstable the classifier's output is for the unlabeled data points under perturbations of the training data. MVAL maintains, what we refer to as, retraining information matrices to keep track of these output scores and exploits two kinds of variance to measure the informativeness and representativeness, respectively. By fusing these variances, MVAL is able to select the instances which are both informative and representative. We employ our technique both in combination with logistic regression and support vector machines and demonstrate that MVAL achieves state-of-the-art performance in experiments on a large number of standard benchmark datasets.
Labeling each instance in a large dataset is extremely labor- and time- consuming . One way to alleviate this problem is active learning, which aims to which discover the most valuable instances for labeling to construct a powerful classifier. Considering both informativeness and representativeness provides a promising way to design a practical active learning. However, most existing active learning methods select instances favoring either informativeness or representativeness. Meanwhile, many are designed based on the binary class, so that they may present suboptimal solutions on the datasets with multiple classes. In this paper, a hybrid informative and representative criterion based multi-class active learning approach is proposed. We combine the informative informativeness and representativeness into one formula, which can be solved under a unified framework. The informativeness is measured by the margin minimum while the representative information is measured by the maximum mean discrepancy. By minimizing the upper bound for the true risk, we generalize the empirical risk minimization principle to the active learning setting. Simultaneously, our proposed method makes full use of the label information, and the proposed active learning is designed based on multiple classes. So the proposed method is not suitable to the binary class but also the multiple classes. We conduct our experiments on twelve benchmark UCI data sets, and the experimental results demonstrate that the proposed method performs better than some state-of-the-art methods.
Multi-armed bandits (MAB) model sequential decision making problems, in which a learner sequentially chooses arms with unknown reward distributions in order to maximize its cumulative reward. Most of the prior work on MAB assumes that the reward distributions of each arm are independent. But in a wide variety of decision problems -- from drug dosage to dynamic pricing -- the expected rewards of different arms are correlated, so that selecting one arm provides information about the expected rewards of other arms as well. We propose and analyze a class of models of such decision problems, which we call {\em global bandits}. In the case in which rewards of all arms are deterministic functions of a single unknown parameter, we construct a greedy policy that achieves {\em bounded regret}, with a bound that depends on the single true parameter of the problem. Hence, this policy selects suboptimal arms only finitely many times with probability one. For this case we also obtain a bound on regret that is {\em independent of the true parameter}; this bound is sub-linear, with an exponent that depends on the informativeness of the arms. We also propose a variant of the greedy policy that achieves $\tilde{\mathcal{O}}(\sqrt{T})$ worst-case and $\mathcal{O}(1)$ parameter dependent regret. Finally, we perform experiments on dynamic pricing and show that the proposed algorithms achieve significant gains with respect to the well-known benchmarks.
In this paper, we consider the approximate weighted graph matching problem and introduce stable and informative first and second order compatibility terms suitable for inclusion into the popular integer quadratic program formulation. Our approach relies on a rigorous analysis of stability of spectral signatures based on the graph Laplacian. In the case of the first order term, we derive an objective function that measures both the stability and informativeness of a given spectral signature. By optimizing this objective, we design new spectral node signatures tuned to a specific graph to be matched. We also introduce the pairwise heat kernel distance as a stable second order compatibility term; we justify its plausibility by showing that in a certain limiting case it converges to the classical adjacency matrix-based second order compatibility function. We have tested our approach on a set of synthetic graphs, the widely-used CMU house sequence, and a set of real images. These experiments show the superior performance of our first and second order compatibility terms as compared with the commonly used ones.