Open-domain conversational QA (ODCQA) calls for effective question rewriting (QR), as the questions in a conversation typically lack proper context for the QA model to interpret. In this paper, we compare two types of QR approaches, generative and expansive QR, in end-to-end ODCQA systems with recently released QReCC and OR-QuAC benchmarks. While it is common practice to apply the same QR approach for both the retriever and the reader in the QA system, our results show such strategy is generally suboptimal and suggest expansive QR is better for the sparse retriever and generative QR is better for the reader. Furthermore, while conversation history modeling with dense representations outperforms QR, we show the advantages to apply both jointly, as QR boosts the performance especially when limited history turns are considered.

We consider the problem of reliably selecting an optimal subset of fixed size from a given set of choice alternatives, based on noisy information about the quality of these alternatives. Problems of similar kind have been tackled by means of adaptive sampling schemes called racing algorithms. However, in contrast to existing approaches, we do not assume that each alternative is characterized by a real-valued random variable, and that samples are taken from the corresponding distributions. Instead, we only assume that alternatives can be compared in terms of pairwise preferences. We propose and formally analyze a general preference-based racing algorithm that we instantiate with three specific ranking procedures and corresponding sampling schemes. Experiments with real and synthetic data are presented to show the efficiency of our approach.

MOTIVATION: Antiretroviral treatment regimens can sufficiently suppress viral replication in human immunodeficiency virus (HIV)-infected patients and prevent the progression of the disease. However, one of the factors contributing to the progression of the disease despite ongoing antiretroviral treatment is the emergence of drug resistance. The high mutation rate of HIV can lead to a fast adaptation of the virus under drug pressure, thus to failure of antiretroviral treatment due to the evolution of drug-resistant variants. Moreover, cross-resistance phenomena have been frequently found in HIV-1, leading to resistance not only against a drug from the current treatment, but also to other not yet applied drugs. Automatic classification and prediction of drug resistance is increasingly important in HIV research as well as in clinical settings, and to this end, machine learning techniques have been widely applied. Nevertheless, cross-resistance information was not taken explicitly into account, yet. RESULTS: In our study, we demonstrated the use of cross-resistance information to predict drug resistance in HIV-1. We tested a set of more than 600 reverse transcriptase sequences and corresponding resistance information for six nucleoside analogues. Based on multilabel classification models and cross-resistance information, we were able to significantly improve overall prediction accuracy for all drugs, compared with single binary classifiers without any additional information. Moreover, we identified drug-specific patterns within the reverse transcriptase sequences that can be used to determine an optimal order of the classifiers within the classifier chains. These patterns are in good agreement with known resistance mutations and support the use of cross-resistance information in such prediction models. CONTACT: [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Several machine learning methods allow for abstaining from uncertain predictions. While being common for settings like conventional classification, abstention has been studied much less in learning to rank. We address abstention for the label ranking setting, allowing the learner to declare certain pairs of labels as being incomparable and, thus, to predict partial instead of total orders. In our method, such predictions are produced via thresholding the probabilities of pairwise preferences between labels, as induced by a predicted probability distribution on the set of all rankings. We formally analyze this approach for the Mallows and the Plackett-Luce model, showing that it produces proper partial orders as predictions and characterizing the expressiveness of the induced class of partial orders. These theoretical results are complemented by experiments demonstrating the practical usefulness of the approach.

Most of the multi-label classification (MLC) methods proposed in recent years intended to exploit, in one way or the other, dependencies between the class labels. Comparing to simple binary relevance learning as a baseline, any gain in performance is normally explained by the fact that this method is ignoring such dependencies. Without questioning the correctness of such studies, one has to admit that a blanket explanation of that kind is hiding many subtle details, and indeed, the underlying mechanisms and true reasons for the improvements reported in experimental studies are rarely laid bare. Rather than proposing yet another MLC algorithm, the aim of this paper is to elaborate more closely on the idea of exploiting label dependence, thereby contributing to a better understanding of MLC. Adopting a statistical perspective, we claim that two types of label dependence should be distinguished, namely conditional and marginal dependence. Subsequently, we present three scenarios in which the exploitation of one of these types of dependence may boost the predictive performance of a classifier. In this regard, a close connection with loss minimization is established, showing that the benefit of exploiting label dependence does also depend on the type of loss to be minimized. Concrete theoretical results are presented for two representative loss functions, namely the Hamming loss and the subset 0/1 loss. In addition, we give an overview of state-of-the-art decomposition algorithms for MLC and we try to reveal the reasons for their effectiveness. Our conclusions are supported by carefully designed experiments on synthetic and benchmark data.

We propose a novel method for preference learning or, more specifically, learning to rank, where the task is to learn a ranking model that takes a subset of alternatives as input and produces a ranking of these alternatives as output. Just like in the case of conventional classifier learning, training information is provided in the form of a set of labeled instances, with labels or, say, preference degrees taken from an ordered categorical scale. This setting is known as multipartite ranking in the literature. Our approach is based on the idea of using the (discrete) Choquet integral as an underlying model for representing ranking functions. Being an established aggregation function in fields such as multiple criteria decision making and information fusion, the Choquet integral offers a number of interesting properties that make it attractive from a machine learning perspective, too. The learning problem itself comes down to properly specifying the fuzzy measure on which the Choquet integral is defined. This problem is formalized as a margin maximization problem and solved by means of a cutting plane algorithm. The performance of our method is tested on a number of benchmark datasets.

The prediction accuracy of any learning algorithm highly depends on the quality of the selected features; but often, the task of feature construction and selection is tedious and nonscalable. In recent years, however, there have been numerous projects with the goal of constructing general-purpose or domain-specific knowledge bases with entity-relationship-entity triples extracted from various Web sources or collected from user communities, e.g. YAGO, DBpedia, Freebase, UMLS, etc. This paper advocates the simple and yet far-reaching idea that the structured knowledge contained in such knowledge bases can be exploited to automatically extract features for general learning tasks. We introduce an expressive graph-based language for extracting features from such knowledge bases and a theoretical framework for constructing feature vectors from the extracted features. Our experimental evaluation on different learning scenarios provides evidence that the features derived through our framework can considerably improve the prediction accuracy, especially when the labeled data at hand is sparse.

The F-measure, originally introduced in information retrieval, is nowadays routinely used as a performance metric for problems such as binary classification, multi-label classification, and structured output prediction. Optimizing this measure remains a statistically and computationally challenging problem, since no closed-form maximizer exists. Current algorithms are approximate and typically rely on additional assumptions regarding the statistical distribution of the binary response variables. In this paper, we present an algorithm which is not only computationally efficient but also exact, regardless of the underlying distribution. The algorithm requires only a quadratic number of parameters of the joint distribution (with respect to the number of binary responses). We illustrate its practical performance by means of experimental results for multi-label classification.

In the database outsourcing paradigm, a data owner (DO) delegates its DBMS administration to a specialized service provider (SP) that receives and processes queries from clients. The traditional outsourcing model (TOM) requires that the DO and the SP maintain authenticated data structures to enable authentication of query results. In this paper, we present SAE, a novel outsourcing model that separates authentication from query execution. Specifically, the DO does not perform any task except for maintaining its dataset (if there are updates). The SP only stores the DO's dataset and computes the query results using a conventional DBMS. All security-related tasks are outsourced to a separate trusted entity (TE), which maintains limited authentication information about the original dataset. A client contacts the TE when it wishes to establish the correctness of a result returned by the SP. The TE efficiently generates a verification token of negligible size. The client can verify the token with minimal cost. SAE eliminates the participation of the DO and the SP in the authentication process, and outperforms TOM in every aspect, including processing cost for all parties involved, communication overhead, query response time and ease of implementation in practical applications.

The label ranking problem consists of learning a model that maps instances to total orders over a finite set of predefined labels. This paper introduces new methods for label ranking that complement and improve upon existing approaches. More specifically, we propose extensions of two methods that have been used extensively for classification and regression so far, namely instance-based learning and decision tree induction. The unifying element of the two methods is a procedure for locally estimating predictive probability models for label rankings.

In data outsourcing model, data owners engage third-party data servers (called publishers) to manage their data and process queries on their behalf. As these publishers may be untrusted or susceptible to attacks, it could produce incorrect query resu