Self-training methods have gained significant attention in recent years due to their effectiveness in leveraging small labeled datasets and large unlabeled observations for prediction tasks. These models identify decision boundaries in low-density regions without additional assumptions about data distribution, using the confidence scores of a learned classifier. The core principle of self-training involves iteratively assigning pseudo-labels to unlabeled samples with confidence scores above a certain threshold, enriching the labeled dataset and retraining the classifier. This paper presents self-training methods for binary and multi-class classification, along with variants and related approaches such as consistency-based methods and transductive learning. We also briefly describe self-supervised learning and reinforced self-training. Furthermore, we highlight popular applications of self-training and discuss the importance of dynamic thresholding and reducing pseudo-label noise for performance improvement. To the best of our knowledge, this is the first thorough and complete survey on self-training.

Pretrained Language Models (PLMs) are the de facto backbone of most state-of-the-art NLP systems. In this paper, we introduce a family of domain-specific pretrained PLMs for French, focusing on three important domains: transcribed speech, medicine, and law. We use a transformer architecture based on efficient methods (LinFormer) to maximise their utility, since these domains often involve processing long documents. We evaluate and compare our models to state-of-the-art models on a diverse set of tasks and datasets, some of which are introduced in this paper. We gather the datasets into a new French-language evaluation benchmark for these three domains. We also compare various training configurations: continued pretraining, pretraining from scratch, as well as single- and multi-domain pretraining. Extensive domain-specific experiments show that it is possible to attain competitive downstream performance even when pre-training with the approximative LinFormer attention mechanism. For full reproducibility, we release the models and pretraining data, as well as contributed datasets.

Machine learning techniques often require large labeled training sets to attain optimal performance. However, acquiring labeled data can pose challenges in practical scenarios. Pool-based active learning methods aim to select the most relevant data points for training from a pool of unlabeled data. Nonetheless, these methods heavily rely on the initial labeled dataset, often chosen randomly. In our study, we introduce a novel approach specifically tailored for multi-class classification tasks, utilizing Proper Topological Regions (PTR) derived from topological data analysis (TDA) to efficiently identify the initial set of points for labeling. Through experiments on various benchmark datasets, we demonstrate the efficacy of our method and its competitive performance compared to traditional approaches, as measured by average balanced classification accuracy.

In recent years, the significance of test logs in ensuring system reliability and diagnosing runtime events has grown significantly, particularly with software expanding into various domains, necessitating rigorous verification and validation processes. However, the complexity and cost of testing have prompted a shift towards automation. This paper addresses the challenges of automated software testing through root-cause event detection. The proposed approach initially involves parsing and partitioning logs, followed by representing test events as dense vectors in a continuous space, enabling the capture of semantic similarities and relationships among events based on their sequence positions. Subsequently, test events are clustered in this embedded space, and each log partition is represented as a vector, with its characteristics reflecting the number of events in the log partition present in the clusters. Through two distinct case studies, we demonstrate that the final clustering of log partitions in this new space efficiently identifies root cause events. We evaluate our approach on two applications and anticipate its contribution as a cornerstone for future research and deployment of automated log mining.

In this paper, we propose a probabilistic framework for analyzing a multi-class majority vote classifier in the case where training data is partially labeled. First, we derive a multi-class transductive bound over the risk of the majority vote classifier, which is based on the classifier's vote distribution over each class. Then, we introduce a mislabeling error model to analyze the error of the majority vote classifier in the case of the pseudo-labeled training data. We derive a generalization bound over the majority vote error when imperfect labels are given, taking into account the mean and the variance of the prediction margin. Finally, we demonstrate an application of the derived transductive bound for self-training to find automatically the confidence threshold used to determine unlabeled examples for pseudo-labeling. Empirical results on different data sets show the effectiveness of our framework compared to several state-of-the-art semi-supervised approaches.

We investigate the generalization properties of a self-training algorithm with halfspaces. The approach learns a list of halfspaces iteratively from labeled and unlabeled training data, in which each iteration consists of two steps: exploration and pruning. In the exploration phase, the halfspace is found sequentially by maximizing the unsigned-margin among unlabeled examples and then assigning pseudo-labels to those that have a distance higher than the current threshold. These pseudo-labels are allegedly corrupted by noise. The training set is then augmented with noisy pseudo-labeled examples, and a new classifier is trained. This process is repeated until no more unlabeled examples remain for pseudo-labeling. In the pruning phase, pseudo-labeled samples that have a distance to the last halfspace greater than the associated unsigned-margin are then discarded. We prove that the misclassification error of the resulting sequence of classifiers is bounded and show that the resulting semi-supervised approach never degrades performance compared to the classifier learned using only the initial labeled training set. Experiments carried out on a variety of benchmarks demonstrate the efficiency of the proposed approach compared to state-of-the-art methods.

In this paper, we propose a Neural Architecture Search strategy based on self supervision and semi-supervised learning for the task of semantic segmentation. Our approach builds an optimized neural network (NN) model for this task by solving a jigsaw pretext problem identified by self-supervised learning over unlabeled training data, and, leveraging the structure of the unlabeled data with semi-supervised learning. Dynamic routing with a gradient descent approach is used to find the architecture of the NN model. Experiments on the Cityscapes and PASCAL VOC 2012 datasets show that the found neural network is four times more efficient than a state-of-the-art hand designed NN model in terms of floating-point operations.

Telemetry data (e.g.: CPU and memory usage) is an essential source of information for a software system that projects the system’s health. Anomalies in telemetry data warn system administrators about an imminent failure or deterioration of service quality. However, input events to the system (such as service requests) are the cause of abnormal system behaviour and, thus, anomalous telemetry data. By observing input events, one might predict anomalies even before they appear in telemetry data, thus giving the system administrator even earlier warning before the failure. Finding a correlation between input events and anomalies in telemetry data is challenging in many cases. This paper proposes a machine learning approach to learn the causality correlation between input event sequences and telemetry data. To this aim, a Natural Language Processing(NLP) approach is employed to create a concept space model to distinguish between normal and abnormal test sequences. Based on a vectorized representation of each input sequence, the concept space indicates whether the sequence will cause a system failure. Since the meaning of fault is not established in system status Telemetry-based fault detection, the suggested technique first detects periods of time when a software system status encounters aberrant situations (Bug-Zones). An extensive study on a real-world database acquired by a telecommunication operator and an open-source microservice software demonstrates that our approach achieves 71% and 90% accuracy as a Bug-Zones predictor.

In this paper, we consider the task of ranking individuals based on the potential benefit of being "treated" (e.g. by a drug or exposure to recommendations or ads), referred to as Uplift Modeling in the literature. This application has gained a surge of interest in recent years and it is found in many applications such as personalized medicine, recommender systems or targeted advertising. In real life scenarios the capacity of models to rank individuals by potential benefit is measured by the Area Under the Uplift Curve (AUUC), a ranking metric related to the well known Area Under ROC Curve. In the case where the objective function, for learning model parameters, is different from AUUC, the capacity of the resulting system to generalize on AUUC is limited. To tackle this issue, we propose to learn a model that directly optimizes an upper bound on AUUC. To find such a model we first develop a generalization bound on AUUC and then derive from it a learning objective called AUUC-max, usable with linear and deep models. We empirically study the tightness of this generalization bound, its effectiveness for hyperparameters tuning and show the efficiency of the proposed learning objective compared to a wide range of competitive baselines on two classical uplift modeling benchmarks using real-world datasets.

In this paper, we propose a theoretically supported sequential strategy for training a large-scale Recommender System (RS) over implicit feedback, mainly in the form of clicks. The proposed approach consists in minimizing pairwise ranking loss over blocks of consecutive items constituted by a sequence of non-clicked items followed by a clicked one for each user. We present two variants of this strategy where model parameters are updated using either the momentum method or a gradient-based approach. To prevent updating the parameters for an abnormally high number of clicks over some targeted items (mainly due to bots), we introduce an upper and a lower threshold on the number of updates for each user. These thresholds are estimated over the distribution of the number of blocks in the training set. They affect the decision of RS by shifting the distribution of items that are shown to the users. Furthermore, we provide a convergence analysis of both algorithms and demonstrate their practical efficiency over six large-scale collections with respect to various ranking measures and computational time.

Kidney Disease (KD) may hide complex causes and is associated with a tremendous socio-economic impact. Timely identification and management from the first level of medical care represent the most effective strategy to address the growing global burden sustainably. Clinical practice guidelines suggest utilizing estimated Glomerular Filtration Rate (eGFR) for routine evaluation within a screening purpose. Accordingly, the analysis of Electronic Health Records (EHRs) using Machine Learning techniques offers great opportunities to monitor and predict the eGFR trend over time. This paper aims to propose a novel Semi-Supervised Multi-Task Learning (SS-MTL) approach for predicting short-term KD evolution on multiple General Practitioners' EHR data. We demonstrated that the SS-MTL approach can (i) capture the eGFR temporal evolution by imposing a temporal relatedness between consecutive time windows and (ii) exploit useful information from unlabeled patients when labeled patients are less numerous with a gain of up to 4.1% in terms of Recall. This situation reflects the real-case scenario, where available labeled samples are limited, but those unlabeled much more abundant. The SS-MTL approach, also given the high level of interpretability, might be the ideal candidate in general practice to get integrated within a decision support system for KD screening purposes.

In this paper, we propose a transductive bound over the risk of the majority vote classifier learned with partially labeled data for the multi-class classification. The bound is obtained by considering the class confusion matrix as an error indicator and it involves the margin distribution of the classifier over each class and a bound over the risk of the associated Gibbs classifier. When this latter bound is tight and, the errors of the majority vote classifier per class are concentrated on a low margin zone; we prove that the bound over the Bayes classifier’ risk is tight. As an application, we extend the self-learning algorithm to the multi-class case. The algorithm iteratively assigns pseudo-labels to a subset of unlabeled training examples that have their associated class margin above a threshold obtained from the proposed transductive bound. Empirical results on different data sets show the effectiveness of our approach compared to the same algorithm where the threshold is fixed manually, to the extension of TSVM to multi-class classification and to a graph-based semi-supervised algorithm.

In this paper, we propose a robust sequential learning strategy for training large-scale Recommender Systems (RS) over implicit feedback mainly in the form of clicks. Our approach relies on the minimization of a pairwise ranking loss over blocks of consecutive items constituted by a sequence of non-clicked items followed by a clicked one for each user. Parameter updates are discarded if for a given user the number of sequential blocks is below or above some given thresholds estimated over the distribution of the number of blocks in the training set. This is to prevent from an abnormal number of clicks over some targeted items, mainly due to bots; or very few user interactions. Both scenarios affect the decision of RS and imply a shift over the distribution of items that are shown to the users. We provide a theoretical analysis showing that in the case where the ranking loss is convex, the deviation between the loss with respect to the sequence of weights found by the proposed algorithm and its minimum is bounded. Furthermore, experimental results on five large-scale collections demonstrate the efficiency of the proposed algorithm with respect to the state-of-the-art approaches, both regarding different ranking measures and computation time.

Distributed learning aims at computing high-quality models by training over scattered data. This covers a diversity of scenarios, including computer clusters or mobile agents. One of the main challenges is then to deal with heterogeneous machines and unreliable communications. In this setting, we propose and analyze a flexible asynchronous optimization algorithm for solving nonsmooth learning problems. Unlike most existing methods, our algorithm is adjustable to various levels of communication costs, machines computational powers, and data distribution evenness. We prove that the algorithm converges linearly with a fixed learning rate that does not depend on communication delays nor on the number of machines. Although long delays in communication may slow down performance, no delay can break convergence.

We propose Rademacher complexity bounds for multi-class classifiers trained with a two-step semi-supervised model. In the first step, the algorithm partitions the partially labeled data and then identifies dense clusters containing k predominant classes using the labeled training examples such that the proportion of their non-predominant classes is below a fixed threshold stands for clustering consistency. In the second step, a classifier is trained by minimizing a margin empirical loss over the labeled training set and a penalization term measuring the disability of the learner to predict the k predominant classes of the identified clusters. The resulting data-dependent generalization error bound involves the margin distribution of the classifier, the stability of the clustering technique used in the first step and Rademacher complexity terms corresponding to partially labeled training data. Our theoretical result exhibit convergence rates extending those proposed in the literature for the binary case, and experimental results on different multi-class classification problems show empirical evidence that supports the theory.

In this paper, we present a novel and publicly available dataset for online recommendation provided by Purch1. The dataset records the clicks generated by users of one of Purch's high-tech website over the ads they have been shown for one month. In addition, the dataset contains contextual information about offers such as offer titles and keywords, as well as the anonymized content of the page on which offers were displayed. Then, besides a detailed description of the dataset, we evaluate the performance of six popular baselines and propose a simple yet effective strategy on how to overcome the existing challenges inherent to implicit feedback and popularity bias introduced while designing an efficient and scalable recommendation algorithm. More specifically, we propose to demonstrate the importance of introducing diversity based on an appropriate representation of items in Recommender Systems, when the available feedback is strongly biased.

We propose Rademacher complexity bounds for multi-class classifiers trained with a two-step semi-supervised model. In the first step, the algorithm partitions the partially labeled data and then identifies dense clusters containing k predominant classes using the labeled training examples such that the proportion of their non-predominant classes is below a fixed threshold stands for clustering consistency. In the second step, a classifier is trained by minimizing a margin empirical loss over the labeled training set and a penalization term measuring the disability of the learner to predict the k predominant classes of the identified clusters. The resulting data-dependent generalization error bound involves the margin distribution of the classifier, the stability of the clustering technique used in the first step and Rademacher complexity terms corresponding to partially labeled training data. Our theoretical result exhibit convergence rates extending those proposed in the literature for the binary case, and experimental results on different multi-class classification problems show empirical evidence that supports the theory.

Social media has become a major source for analyzing all aspects of daily life. Thanks to dedicated latent topic analysis methods such as the Ailment Topic Aspect Model (ATAM), public health can now be observed on Twitter. In this work, we are interested in using social media to monitor people's health overtime. The use of tweets has several benefits including instantaneous data availability at virtually no cost. Early monitoring of health data is complementary to post-factum studies and enables a range of applications such as measuring behavioral risk factors and triggering health campaigns. We formulate two problems: health transition detection and health transition prediction. We first propose the Temporal Ailment Topic Aspect Model (TM-ATAM), a new latent model dedicated to solving the first problem by capturing transitions that involve health-related topics. TM-ATAM is a non-obvious extension to ATAM that was designed to extract health-related topics. It learns health-related topic transitions by minimizing the prediction error on topic distributions between consecutive posts at different time and geographic granularities. To solve the second problem, we develop T-ATAM, a Temporal Ailment Topic Aspect Model where time is treated as a random variable natively inside ATAM. Our experiments on an 8-month corpus of tweets show that TM-ATAM outperforms TM-LDA in estimating health-related transitions from tweets for different geographic populations. We examine the ability of TM-ATAM to detect transitions due to climate conditions in different geographic regions. We then show how T-ATAM can be used to predict the most important transition and additionally compare T-ATAM with CDC (Center for Disease Control) data and Google Flu Trends.

Hesam Amoualian, Wei Lu, Eric Gaussier, Georgios Balikas, Massih R. Amini, Marianne Clausel. Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2017.

We address the problem of multi-class classification in the case where the number of classes is very large. We propose a double sampling strategy on top of a multi-class to binary reduction strategy, which transforms the original multi-class problem into a binary classification problem over pairs of examples. The aim of the sampling strategy is to overcome the curse of long-tailed class distributions exhibited in majority of large-scale multi-class classification problems and to reduce the number of pairs of examples in the expanded data. We show that this strategy does not alter the consistency of the empirical risk minimization principle defined over the double sample reduction. Experiments are carried out on DMOZ and Wikipedia collections with 10,000 to 100,000 classes where we show the efficiency of the proposed approach in terms of training and prediction time, memory consumption, and predictive performance with respect to state-of-the-art approaches.

Traditional sentiment analysis approaches tackle problems like ternary (3-category) and fine-grained (5-category) classification by learning the tasks separately. We argue that such classification tasks are correlated and we propose a multitask approach based on a recurrent neural network that benefits by jointly learning them. Our study demonstrates the potential of multitask models on this type of problems and improves the state-of-the-art results in the fine-grained sentiment classification problem.

In this paper, we describe a novel, publicly available collection for recommendation systems that records the behavior of customers of the European leader in eCommerce advertising, Kelkoo\footnote{\url{https://www.kelkoo.com/}}, during one month. This dataset gathers implicit feedback, in form of clicks, of users that have interacted with over 56 million offers displayed by Kelkoo, along with a rich set of contextual features regarding both customers and offers. In conjunction with a detailed description of the dataset, we show the performance of six state-of-the-art recommender models and raise some questions on how to encompass the existing contextual information in the system.

The exchangeability assumption in topic models like Latent Dirichlet Allocation (LDA) often results in inferring inconsistent topics for the words of text spans like noun-phrases, which are usually expected to be topically coherent. We propose copulaLDA, that extends LDA by integrating part of the text structure to the model and relaxes the conditional independence assumption between the word-specific latent topics given the per-document topic distributions. To this end, we assume that the words of text spans like noun-phrases are topically bound and we model this dependence with copulas. We demonstrate empirically the effectiveness of copulaLDA on both intrinsic and extrinsic evaluation tasks on several publicly available corpora.

We propose in this paper two new models for modeling topic and word-topic dependencies between consecutive documents in document streams. The first model is a direct extension of Latent Dirichlet Allocation model (LDA) and makes use of a Dirichlet distribution to balance the influence of the LDA prior parameters wrt to topic and word-topic distribution of the previous document. The second extension makes use of copulas, which constitute a generic tools to model dependencies between random variables. We rely here on Archimedean copulas, and more precisely on Franck copulas, as they are symmetric and associative and are thus appropriate for exchangeable random variables. Our experiments, conducted on three standard collections that have been used in several studies on topic modeling, show that our proposals outperform previous ones (as dynamic topic models and temporal \LDA), both in terms of perplexity and for tracking similar topics in a document stream.

We introduce an asynchronous distributed stochastic gradient algorithm for matrix factorization based collaborative filtering. The main idea of this approach is to distribute the user-rating matrix across different machines, each having access only to a part of the information, and to asynchronously propagate the updates of the stochastic gradient optimization across the network. Each time a machine receives a parameter vector, it averages its current parameter vector with the received one, and continues its iterations from this new point. Additionally, we introduce a similarity based regularization that constrains the user and item factors to be close to the average factors of their similar users and items found on subparts of the distributed user-rating matrix. We analyze the impact of the regularization terms on MovieLens (100K, 1M, 10M) and NetFlix datasets and show that it leads to a more efficient matrix factorization in terms of Root Mean Square Error (RMSE) and Mean Absolute Error (MAE), and that the asynchronous distributed approach significantly improves in convergence time as compared to an equivalent synchronous distributed approach.

Probabilistic topic models are generative models that describe the content of documents by discovering the latent topics underlying them. However, the structure of the textual input, and for instance the grouping of words in coherent text spans such as sentences, contains much information which is generally lost with these models. In this paper, we propose sentenceLDA, an extension of LDA whose goal is to overcome this limitation by incorporating the structure of the text in the generative and inference processes. We illustrate the advantages of sentenceLDA by comparing it with LDA using both intrinsic (perplexity) and extrinsic (text classification) evaluation tasks on different text collections.

Social media has become a major source for analyzing all aspects of daily life. Thanks to dedicated latent topic analysis methods such as the Ailment Topic Aspect Model (ATAM), public health can now be observed on Twitter. In this work, we are interested in monitoring people's health over time. Recently, Temporal-LDA (TM?LDA) was proposed for efficiently modeling general-purpose topic transitions over time. In this paper, we propose Temporal Ailment Topic Aspect (TM?ATAM), a new latent model dedicated to capturing transitions that involve health-related topics. TM?ATAM learns topic transition parameters by minimizing the prediction error on topic distributions between consecutive posts at different time and geographic granularities. Our experiments on an 8-month corpus of tweets show that it largely outperforms its predecessors.

In this paper, we study flat and hierarchical classification strategies in the context of large-scale taxonomies. Addressing the problem from a learning-theoretic point of view, we first propose a multi-class, hierarchical data dependent bound on the generalization error of classifiers deployed in large-scale taxonomies. This bound provides an explanation to several empirical results reported in the literature, related to the performance of flat and hierarchical classifiers. Based on this bound, we also propose a technique for modifying a given taxonomy through pruning, that leads to a lower value of the upper bound as compared to the original taxonomy. We then present another method for hierarchy pruning by studying approximation error of a family of classifiers, and derive from it features used in a meta-classifier to decide which nodes to prune. We finally illustrate the theoretical developments through several experiments conducted on two widely used taxonomies.

We provide new concentration inequalities for functions of dependent variables. The work extends that of Janson (2004), which proposes concentration inequalities using a combination of the Laplace transform and the idea of fractional graph coloring, as well as many works that derive concentration inequalities using the entropy method (see, e.g., (Boucheron et al., 2003)). We give inequalities for fractionally sub-additive and fractionally self-bounding functions. In the way, we prove a new Talagrand concentration inequality for fractionally sub-additive functions of dependent variables. The results allow us to envision the derivation of generalization bounds for various applications where dependent variables naturally appear, such as in bipartite ranking.

For large-scale category systems, such as Directory Mozilla, which consist of tens of thousand categories, it has been empirically verified in earlier studies that the distribution of documents among categories can be modeled as a power-law distribution. It implies that a significant fraction of categories, referred to as rare categories, have very few documents assigned to them. This characteristic of the data makes it harder for learning algorithms to learn effective decision boundaries which can correctly detect such categories in the test set. In this work, we exploit the distribution of documents among categories to (i) derive an upper bound on the accuracy of any classifier, and (ii) propose a ranking-based algorithm which aims to maximize this upper bound. The empirical evaluation on publicly available large-scale datasets demonstrate that the proposed method not only achieves higher accuracy but also much higher coverage of rare categories as compared to state-of-the-art methods.

This paper provides an overview of the workshop Web-Scale Classification: Web Classification in the Big Data Era which was held in New York City, on February 28th as a workshop of the seventh International Conference on Web Search and Data Mining. The goal of the workshop was to discuss and assess recent research focusing on classification and mining in Web-scale category systems. The workshop brought together members of several communities such web mining, machine learning, text classification and social media mining.

We investigate the problem of learning an IR function on a collection without relevance judgements (called target collection) by transferring knowledge from a selected source collection with relevance judgements. To do so, we first construct, for each query in the target collection, relative relevance judgment pairs using information from the source collection closest to the query (selection and transfer steps), and then learn an IR function from the obtained pairs in the target collection (self-learning step). For the transfer step, the relevance information in the source collection is summarized as a grid that provides, for each term frequency and document frequency values of a word in a document, an empirical estimate of the relevance of the document. The self-learning step iteratively assigns pairwise preferences to documents in the target collection using the scores of the former learned function. We show the effectiveness of our approach through a series of extensive experiments on CLEF and several collections from TREC used either as target or source datasets. Our experiments show the importance of selecting the source collection prior to transfer information to the target collection, and demonstrate that the proposed approach yields results consistently and significantly above state-of-the-art IR functions.

Most photo sharing sites give their users the opportunity to manually label images. The labels collected that way are usually very incomplete due to the size of the image collections: most images are not labeled according to all the categories they belong to, and, conversely, many class have relatively few representative examples. Automated image systems that can deal with small amounts of labeled examples and unbalanced classes are thus necessary to better organize and annotate images. In this work, we propose a multiview semi-supervised bipartite ranking model which allows to leverage the information contained in unlabeled sets of images in order to improve the prediction performance, using multiple descriptions, or views of images. For each topic class, our approach first learns as many view-specific rankers as available views using the labeled data only. These rankers are then improved iteratively by adding pseudo-labeled pairs of examples on which all view-specific rankers agree over the ranking of examples within these pairs. We report on experiments carried out on the NUS-WIDE dataset, which show that the multiview ranking process improves predictive performances when a small number of labeled examples is available specially for unbalanced classes. We show also that our approach achieves significant improvements over a state-of-the art semi-supervised multiview classification model.

We study in this paper flat and hierarchical classification strategies in the context of large-scale taxonomies. To this end, we first propose a multiclass, hierarchical data dependent bound on the generalization error of classifiers deployed in large-scale taxonomies. This bound provides an explanation to several empirical results reported in the literature, related to the performance of flat and hierarchical classifiers. We then introduce another type of bounds targeting the approximation error of a family of classifiers, and derive from it features used in a meta-classifier to decide which nodes to prune (or flatten) in a large-scale taxonomy. We finally illustrate the theoretical developments through several experiments conducted on two widely used taxonomies.

Multiview learning has been shown to be a natural and efficient framework for supervised or semi-supervised learning of multilingual document categorizers. The state-of-the-art co-regularization approach relies on alternate minimizations of a combination of language-specific categorization errors and a disagreement between the outputs of the monolingual text categorizers. This is typically solved by repeatedly training categorizers on each language with the appropriate regularizer. We extend and improve this approach by introducing an on-line learning scheme, where language-specific updates are interleaved in order to iteratively optimize the global cost in one pass. Our experimental results show that this produces similar performance as the batch approach, at a fraction of the computational cost.

Multidocument summarization (MDS) aims for each given query to extract compressed and relevant information with respect to the different query‐related themes present in a set of documents. Many approaches operate in two steps. Themes are first identified from the set, and then a summary is formed by extracting salient sentences within the different documents of each of the identified themes. Among these approaches, latent semantic analysis (LSA) based approaches rely on spectral decomposition techniques to identify the themes. In this article, we propose a major extension of these techniques that relies on the quantum information access (QIA) framework. The latter is a framework developed for modeling information access based on the probabilistic formalism of quantum physics. The QIA framework not only points out the limitations of the current LSA‐based approaches, but motivates a new principled criterium to tackle multidocument summarization that addresses these limitations. As a byproduct, it also provides a way to enhance the LSA‐based approaches. Extensive experiments on the DUC 2005, 2006 and 2007 datasets show that the proposed approach consistently improves over both the LSA‐based approaches and the systems that competed in the yearly DUC competitions. This demonstrates the potential impact of quantum‐inspired approaches to information access in general, and of the QIA framework in particular.

We propose a new method for query-biased multi-document summarization, based on sentence extraction. The summary of multiple documents is created in two steps. Sentences are first clustered; where each cluster corresponds to one of the main themes present in the collection. Inside each theme, sentences are then ranked using a transductive learning-to-rank algorithm based on RankNet, in order to better identify those which are relevant to the query. The final summary contains the top-ranked sentences of each theme. Our approach is validated on DUC 2006 and DUC 2007 datasets.

We investigate the problem of learning document classifiers in a multilingual setting, from collections where labels are only partially available. We address this problem in the framework of multiview learning, where different languages correspond to different views of the same document, combined with semi-supervised learning in order to benefit from unlabeled documents. We rely on two techniques, coregularization and consensus-based self-training, that combine multiview and semi-supervised learning in different ways. Our approach trains different monolingual classifiers on each of the views, such that the classifiers' decisions over a set of unlabeled examples are in agreement as much as possible, and iteratively labels new examples from another unlabeled training set based on a consensus across language-specific classifiers. We derive a boosting-based training algorithm for this task, and analyze the impact of the number of views on the semi-supervised learning results on a multilingual extension of the Reuters RCV1/RCV2 corpus using five different languages. Our experiments show that coregularization and consensus-based self-training are complementary and that their combination is especially effective in the interesting and very common situation where there are few views (languages) and few labeled documents available.

We propose a new multi-view clustering method which uses clustering results obtained on each view as a voting pattern in order to construct a new set of multi-view clusters. Our experiments on a multilingual corpus of documents show that performance increases significantly over simple concatenation and another multi-view clustering technique.

We address the problem of learning classifiers when observations have multiple views, some of which may not be observed for all examples. We assume the existence of view generating functions which may complete the missing views in an approximate way. This situation corresponds for example to learning text classifiers from multilingual collections where documents are not available in all languages. In that case, Machine Translation (MT) systems may be used to translate each document in the missing languages. We derive a generalization error bound for classifiers learned on examples with multiple artificially created views. Our result uncovers a trade-off between the size of the training set, the number of views, and the quality of the view generating functions. As a consequence, we identify situations where it is more interesting to use multiple views for learning instead of classical single view learning. An extension of this framework is a natural way to leverage unlabeled multi-view data in semi-supervised learning. Experimental results on a subset of the Reuters RCV1/RCV2 collections support our findings by showing that additional views obtained from MT may significantly improve the classification performance in the cases identified by our trade-off.

This paper presents a transductive approach to learn ranking functions for extractive multi-document summarization. At the first stage, the proposed approach identifies topic themes within a document collection, which help to identify two sets of relevant and irrelevant sentences to a question. It then iteratively trains a ranking function over these two sets of sentences by optimizing a ranking loss and fitting a prior model built on keywords. The output of the function is used to find further relevant and irrelevant sentences. This process is repeated until a desired stopping criterion is met.

In this paper we propose an extension of the PLSA model in which an extra latent variable allows the model to co-cluster documents and terms simultaneously. We show on three datasets that our extended model produces statistically significant improvements with respect to two clustering measures over the original PLSA and the multinomial mixture MM models.

In this paper we present two transductive bounds on the risk of the majority vote estimated over partially labeled training sets. Our first bound is tight when the additional unlabeled training data are used in the cases where the voted classifier makes its errors on low margin observations and where the errors of the associated Gibbs classifier can accurately be estimated. In semi-supervised learning, considering the margin as an indicator of confidence constitutes the working hypothesis of algorithms which search the decision boundary on low density regions. In this case, we propose a second bound on the joint probability that the voted classifier makes an error over an example having its margin over a fixed threshold. As an application we are interested on self-learning algorithms which assign iteratively pseudo-labels to unlabeled training examples having margin above a threshold obtained from this bound. Empirical results on different datasets show the effectiveness of our approach compared to the same algorithm and the TSVM in which the threshold is fixed manually.

This paper presents a boosting based algorithm for learning a bipartite ranking function (BRF) with partially labeled data. Until now different attempts had been made to build a BRF in a transductive setting, in which the test points are given to the methods in advance as unlabeled data. The proposed approach is a semi-supervised inductive ranking algorithm which, as opposed to transductive algorithms, is able to infer an ordering on new examples that were not used for its training. We evaluate our approach using the TREC-9 Ohsumed and the Reuters-21578 data collections, comparing against two semi-supervised classification algorithms for ROCArea (AUC), uninterpolated average precision (AUP), mean [email protected] (TP) and Precision-Recall (PR) curves. In the most interesting cases where there are an unbalanced number of irrelevant examples over relevant ones, we show our method to produce statistically significant improvements with respect to these ranking measures.

Documents formatted in eXtensible Markup Language (XML) are becoming increasingly available in collections of various document types. In this paper, we present an approach for the summarisation of XML documents. The novelty of this approach lies in that it is based on features not only from the content of documents, but also from their logical structure. We follow a machine learning like, sentence extraction-based summarisation technique. To find which features are more effective for producing summaries this approach views sentence extraction as an ordering task. We evaluated our summarisation model using the INEX dataset. The results demonstrate that the inclusion of features from the logical structure of documents increases the effectiveness of the summariser, and that the learnable system is also effective and well-suited to the task of summarisation in the context of XML documents.

In this paper we propose a general framework to study the generalization properties of binary classifiers trained with data which may be depen- dent, but are deterministically generated upon a sample of independent examples. It provides generalization bounds for binary classification and some cases of ranking problems, and clarifies the relationship between these learning tasks.

With the huge amount of information available electronically, there is an increasing demand for automatic text summarization systems. The use of machine learning techniques for this task allows one to adapt summaries to the user needs and to the corpus characteristics. These desirable properties have motivated an increasing amount of work in this field over the last few years. Most approaches attempt to generate summaries by extracting sentence segments and adopt the supervised learning paradigm which requires to label documents at the text span level. This is a costly process, which puts strong limitations on the applicability of these methods. We investigate here the use of semi-supervised algorithms for summarization. These techniques make use of few labeled data together with a larger amount of unlabeled data. We propose new semi-supervised algorithms for training classification models for text summarization. We analyze their performances on two data sets - the Reuters news-wire corpus and the Computation and Language (cmp_lg) collection of TIPSTER SUMMAC. We perform comparisons with a baseline - non learning - system, and a reference trainable summarizer system.