This Website contains a short introduction to Noisy Data together with the more relevant bibliography and it also contains the complementary material to the SCI2S research group papers on Noisy Data in Data Mining.

The most entertaining video we posted on Video Friday a couple weeks ago was almost certainly Robot Drone Man, a parody of this PPAP (Pen Pineapple Apple Pen) video, which for some reason has 150 million views on YouTube. Parody or not, Robot Drone Man actually exists, and it's a project of Ilhan Bae, a researcher and futurist at the Korea Advanced Institute of Science and Technology (KAIST), who wrote in to tell us about it. Robot Drone Man is an avatar drone, in the same category as other telepresence robots like Double and Beam. It allows a remote human to have an embodied physical presence through a mobile robot, although in this case, the robot can fly, since most of it is a DJI S1000 octocopter. With a height of 1.4 meters (landed), it's designed to match the eye level of people interacting with it, and the remote operator can "gesticulate with two hands and head as if a distant operator exists in person," says Bae, adding that this is "the first trial to couple a telepresence robot in an upright position and drone platform into one body."

Alan Turing Institute lends support to'Commission on Artificial Intelligence' published by the Science and Technology Committee. The establishment of an AI ethics board in the UK has taken a big step forward, with the Alan Turing Institute agreeing to work with the UK government to explore the ethics questions surrounding the development of artificial development. In a letter to the Science and Technology Committee, the Alan Turing Institute welcomed the Committee's recent Report on Robotics and Artificial Intelligence and put itself forward as an institution prepared to take a leading role in taking AI forward. The letter to Committee chair Stephen Metcalfe was in response to a report published by the Committee on 12 October 2016, in which it recommended that a'standing Commission on Artificial Intelligence' be established at the Alan Turing Institute to examine the social, ethical and legal implications of recent and potential developments in AI. "Your Report recommends that a standing Commission on Artificial Intelligence be Should this recommendation be taken forward, we would very much welcome the opportunity to lead the creation of the Commission." MP Stephen Metcalfe returned support in kind, saying in response to the Institute's letter: "We welcome the Alan Turing Institute's support for our report on Robotics and Artificial Intelligence and are pleased that, as the UK's new data science research institute, it is ready to lead the standing Commission on Artificial Intelligence that we recommended establishing" Debate surrounding the ethics concerned with AI has been gathering speed in recent times, with Melanie Mitchell recently telling CBR that the AI community is'not very well prepared' when it comes to the ethical issues that come with using AI in life-critical areas.

Continuous optimization is an important problem in many areas of AI, including vision, robotics, probabilistic inference, and machine learning. Unfortunately, most real-world optimization problems are nonconvex, causing standard convex techniques to find only local optima, even with extensions like random restarts and simulated annealing. We observe that, in many cases, the local modes of the objective function have combinatorial structure, and thus ideas from combinatorial optimization can be brought to bear. Based on this, we propose a problem-decomposition approach to nonconvex optimization. Similarly to DPLL-style SAT solvers and recursive conditioning in probabilistic inference, our algorithm, RDIS, recursively sets variables so as to simplify and decompose the objective function into approximately independent sub-functions, until the remaining functions are simple enough to be optimized by standard techniques like gradient descent. The variables to set are chosen by graph partitioning, ensuring decomposition whenever possible. We show analytically that RDIS can solve a broad class of nonconvex optimization problems exponentially faster than gradient descent with random restarts. Experimentally, RDIS outperforms standard techniques on problems like structure from motion and protein folding.

The interpretation of Large Hadron Collider (LHC) data in the framework of Beyond the Standard Model (BSM) theories is hampered by the need to run computationally expensive event generators and detector simulators. Performing statistically convergent scans of high-dimensional BSM theories is consequently challenging, and in practice unfeasible for very high-dimensional BSM theories. We present here a new machine learning method that accelerates the interpretation of LHC data, by learning the relationship between BSM theory parameters and data. As a proof-of-concept, we demonstrate that this technique accurately predicts natural SUSY signal events in two signal regions at the High Luminosity LHC, up to four orders of magnitude faster than standard techniques. The new approach makes it possible to rapidly and accurately reconstruct the theory parameters of complex BSM theories, should an excess in the data be discovered at the LHC.

We introduce a doubly stochastic proximal gradient algorithm for optimizing a finite average of smooth convex functions, whose gradients depend on numerically expensive expectations. Our main motivation is the acceleration of the optimization of the regularized Cox partial-likelihood (the core model used in survival analysis), but our algorithm can be used in different settings as well. The proposed algorithm is doubly stochastic in the sense that gradient steps are done using stochastic gradient descent (SGD) with variance reduction, where the inner expectations are approximated by a Monte-Carlo Markov-Chain (MCMC) algorithm. We derive conditions on the MCMC number of iterations guaranteeing convergence, and obtain a linear rate of convergence under strong convexity and a sublinear rate without this assumption. We illustrate the fact that our algorithm improves the state-of-the-art solver for regularized Cox partial-likelihood on several datasets from survival analysis.

Variable selection for models including interactions between explanatory variables often needs to obey certain hierarchical constraints. The weak or strong structural hierarchy requires that the existence of an interaction term implies at least one or both associated main effects to be present in the model. Lately, this problem has attracted a lot of attention, but existing computational algorithms converge slow even with a moderate number of predictors. Moreover, in contrast to the rich literature on ordinary variable selection, there is a lack of statistical theory to show reasonably low error rates of hierarchical variable selection. This work investigates a new class of estimators that make use of multiple group penalties to capture structural parsimony. We give the minimax lower bounds for strong and weak hierarchical variable selection and show that the proposed estimators enjoy sharp rate oracle inequalities. A general-purpose algorithm is developed with guaranteed convergence and global optimality. Simulations and real data experiments demonstrate the efficiency and efficacy of the proposed approach.

Although the Lasso has been extensively studied, the relationship between its prediction performance and the correlations of the covariates is not fully understood. In this paper, we give new insights into this relationship in the context of multiple linear regression. We show, in particular, that the incorporation of a simple correlation measure into the tuning parameter can lead to a nearly optimal prediction performance of the Lasso even for highly correlated covariates. However, we also reveal that for moderately correlated covariates, the prediction performance of the Lasso can be mediocre irrespective of the choice of the tuning parameter. We finally show that our results also lead to near-optimal rates for the least-squares estimator with total variation penalty.

A major challenge in crowdsourcing evaluation tasks like labeling objects, grading assignments in online courses, etc., is that of eliciting truthful responses from agents in the absence of verifiability. In this paper, we propose new reward mechanisms for such settings that, unlike many previously studied mechanisms, impose minimal assumptions on the structure and knowledge of the underlying generating model, can account for heterogeneity in the agents' abilities, require no extraneous elicitation from them, and furthermore allow their beliefs to be (almost) arbitrary. These mechanisms have the simple and intuitive structure of an output agreement mechanism: an agent gets a reward if her evaluation matches that of her peer, but unlike the classic output agreement mechanism, this reward is not the same across evaluations, but is inversely proportional to an appropriately defined popularity index of each evaluation. The popularity indices are computed by leveraging the existence of a large number of similar tasks, which is a typical characteristic of these settings. Experiments performed on MTurk workers demonstrate higher efficacy (with a $p$-value of $0.02$) of these mechanisms in inducing truthful behavior compared to the state of the art.

To better understand ISIS drones, I spoke with an investigator at Conflict Armament research, who requested anonymity given the sensitive nature of the work. When the investigator entered the workshop, there were no completed drones inside. Instead, they saw plywood fuselages and styrofoam wings, as well as a missile from a man-portable anti-air defense system, or MANPADS. "For us it implied that they were trying to arm it, arm their drones with something that would be light enough to be carried by a drone, but also that would have the right kind of explosives for potency," they said. Many of their finding were published in a report on the Islamic State's Weaponized Drones.