In this tutorial, we present a portion of unique industry experience in efficient data labeling via crowdsourcing shared by both leading researchers and engineers from Yandex. We will make an introduction to data labeling via public crowdsourcing marketplaces and will present the key components of efficient label collection. This will be followed by a practice session, where participants will choose one of the real label collection tasks, experiment with selecting settings for the labeling process, and launch their label collection project on one of the largest crowdsourcing marketplaces. The projects will be run on real crowds within the tutorial session. While the crowd performers are annotating the project set up by the attendees, we will present the major theoretical results in efficient aggregation, incremental relabeling, and dynamic pricing. We will also discuss their strengths and weaknesses as well as applicability to real-world tasks, summarizing our five year-long research and industrial expertise in crowdsourcing. Finally, participants will receive a feedback about their projects and practical advice on how to make them more efficient. We invite beginners, advanced specialists, and researchers to learn how to collect high quality labeled data and do it efficiently.

In this tutorial, we present a portion of unique industry experience in efficient data labelling via crowdsourcing shared by both leading researchers and engineers from Yandex. We will make an introduction to data labelling via public crowdsourcing marketplaces and will present key components of efficient label collection. This will be followed by a practice session, where participants will choose one of the real label collection tasks, experiment with selecting settings for the labelling process, and launch their label collection project on Yandex.Toloka, one of the largest crowdsourcing marketplaces. The projects will be run on real crowds within the tutorial session. Finally, participants will receive a feedback about their projects and practical advice to make them more efficient. We expect that our tutorial will address an audience with a wide range of background and interests. We do not require specific prerequisite knowledge or skills. We invite beginners, advanced specialists, and researchers to learn how to efficiently collect labelled data.

Internet companies use crowdsourcing to collect large amounts of data needed for creating products based on machine learning techniques. A significant source of such labels for OCR data sets is (re)CAPTCHA, which distinguishes humans from automated bots by asking them to recognize text and, at the same time, receives new labeled data in this way. An important component of such approach to data collection is the reduction of noisy labels produced by bots and non-qualified users.

Labeled training data is an essential part of any supervised machine learning framework. In practice, there is a trade-off between the quality of a label and its cost. In this paper, we consider a problem of learning to rank on a large-scale dataset with low-quality relevance labels aiming at maximizing the quality of a trained ranker on a small validation dataset with high-quality ground truth relevance labels. Motivated by the classical Gauss-Markov theorem for the linear regression problem, we formulate the problems of (1) reweighting training instances and (2) remapping learning targets. We propose meta–gradient decision tree learning framework for optimizing weight and target functions by applying gradient-based hyperparameter optimization. Experiments on a large-scale real-world dataset demonstrate that we can significantly improve state-of-the-art machine-learning algorithms by incorporating our framework.

Relevance labels is the essential part of any learning to rank framework. The rapid development of crowdsourcing platforms led to a significant reduction of the cost of manual labeling. This makes it possible to collect very large sets of labeled documents to train a ranking algorithm. However, relevance labels acquired via crowdsourcing are typically coarse and noisy, so certain consensus models are used to measure the quality of labels and to reduce the noise. This noise is likely to affect a ranker trained on such labels, and, since none of the existing consensus models directly optimizes ranking quality, one has to apply some heuristics to utilize the output of a consensus model in a ranking algorithm, e.g., to use majority voting among workers to get consensus labels. The major goal of this paper is to unify existing approaches to consensus modeling and noise reduction within a learning to rank framework. Namely, we present a machine learning algorithm aimed at improving the performance of a ranker trained on a crowdsourced dataset by proper remapping of labels and reweighting of samples. In the experimental part, we use several characteristics of workers/labels extracted via various consensus models in order to learn the remapping and reweighting functions. Our experiments on a large-scale dataset demonstrate that we can significantly improve state-of-the-art machine-learning algorithms by incorporating our framework.

This paper studies theoretically and empirically a method of turning machine-learning algorithms into probabilistic predictors that automatically enjoys a property of validity (perfect calibration) and is computationally efficient. The price to pay for perfect calibration is that these probabilistic predictors produce imprecise (in practice, almost precise for large data sets) probabilities. When these imprecise probabilities are merged into precise probabilities, the resulting predictors, while losing the theoretical property of perfect calibration, are consistently more accurate than the existing methods in empirical studies.

A certificate of non-membership for a Boolean function f with respect to a class 𝒞, f ∉ 𝒞, is a set S of input strings such that the values of f on strings from S are inconsistent with any function h ∈ 𝒞. We study certificates of non-membership wi