Branislav Kveton

Solving factored MDPs with hybrid state and action variables

Efficient representations and solutions for large decision problems with continuous and discrete variables are among the most important challenges faced by the designers of automated decision support systems. In this paper, we describe a novel hybrid factored Markov decision process (MDP) model that allows for a compact representation of these problems, and a new hybrid approximate linear programming (HALP) framework that permits their efficient solutions. The central idea of HALP is to approximate the optimal value function by a linear combination of basis functions and optimize its weights by linear programming. We analyze both theoretical and computational aspects of this approach, and demonstrate its scale-up potential on several hybrid optimization problems.

How to hide the elephant- or the donkey- in the room: Practical privacy against statistical inference for large data

We propose a practical methodology to protect a users private data, when he wishes to publicly release data that is correlated with his private data, in the hope of getting some utility. Our approach relies on a general statistical inference framework that captures the privacy threat under inference attacks, given utility constraints. Under this framework, data is distorted before it is released, according to a privacy-preserving probabilistic mapping. This mapping is obtained by solving a convex optimization problem, which minimizes information leakage under a distortion constraint. We address a practical challenge encountered when applying this theoretical framework to real world data: the optimization may become untractable and face scalability issues when data assumes values in large size alphabets, or is high dimensional. Our work makes two major contributions. We first reduce the optimization size by introducing a quantization step, and show how to generate privacy mappings under quantization. Second, we evaluate our method on a dataset showing correlations between political views and TV viewing habits, and demonstrate that good privacy properties can be achieved with limited distortion so as not to undermine the original purpose of the publicly released data, e.g. recommendations.

Inferring Identity Using Accelerometers in Television Remote Controls

We show that accelerometers embedded in a television remote control can be used to distinguish household members based on the unique way each person wields the remote. This personalization capability can be applied to enhance digital video recorders with show recommendations per family-member instead of per device or as an enabling technology for targeted advertising. Based on five 1-3 week data sets collected from real homes, using 372 features including key press codes, key press timing, and 3-axis acceleration parameters including dominant frequency, energy, mean, and variance, we show household member identification accuracy of 70-92% with a Max-Margin Markov Network (M3N) classifier.

Managing Your Private and Public Data: Bringing Down Inference Attacks Against Your Privacy

We propose a practical methodology to protect a users private data, when he wishes to publicly release data that is correlated with his private data, to get some utility. Our approach relies on a general statistical inference framework that captures the privacy threat under inference attacks, given utility constraints. Under this framework, data is distorted before it is released, according to a probabilistic privacy mapping. This mapping is obtained by solving a convex optimization problem, which minimizes information leakage under a distortion constraint. We address practical challenges encountered when applying this theoretical framework to real world data. On one hand, the design of optimal privacy mappings requires knowledge of the prior distribution linking private data and data to be released, which is often unavailable in practice. On the other hand, the optimization may become untractable when data assumes values in large size alphabets, or is high dimensional. Our work makes three major contributions. First, we provide bounds on the impact of a mismatched prior on the privacy-utility tradeoff. Second, we show how to reduce the optimization size by introducing a quantization step, and how to generate privacy mappings under quantization. Third, we evaluate our method on two datasets, including a new dataset that we collected, showing correlations between political convictions and TV viewing habits. We demonstrate that good privacy properties can be achieved with limited distortion so as not to undermine the original purpose of the publicly released data, e.g., recommendations.

Conditional Anomaly Detection with Soft Harmonic Functions

In this paper, we consider the problem of conditional anomaly detection that aims to identify data instances with an unusual response or a class label. We develop a new non-parametric approach for conditional anomaly detection based on the soft harmonic solution, with which we estimate the confidence of the label to detect anomalous mislabeling. We further regularize the solution to avoid the detection of isolated examples and examples on the boundary of the distribution support. We demonstrate the efficacy of the proposed method on several synthetic and UCI ML datasets in detecting unusual labels when compared to several baseline approaches. We also evaluate the performance of our method on a real-world electronic health record dataset where we seek to identify unusual patient-management decisions.

Online semi-supervised perception: Real-time learning without explicit feedback

This paper proposes an algorithm for real-time learning without explicit feedback. The algorithm combines the ideas of semi-supervised learning on graphs and online learning. In particular, it iteratively builds a graphical representation of its world and updates it with observed examples. Labeled examples constitute the initial bias of the algorithm and are provided offline, and a stream of unlabeled examples is collected online to update this bias. We motivate the algorithm, discuss how to implement it efficiently, prove a regret bound on the quality of its solutions, and apply it to the problem of real-time face recognition. Our recognizer runs in real time, and achieves superior precision and recall on 3 challenging video datasets.

Minimal Interaction Search in Recommender Systems

While numerous works study algorithms for predicting item ratings in recommender systems, the area of the user-recommender interaction remains largely under-explored. In this work, we look into user interaction with the recommendation list, aiming to devise a method that allows users to discover items of interest in a minimal number of interactions. We propose generalized linear search (GLS), a combination of linear and generalized searches that brings together the benefits of both approaches. We prove that GLS performs at least as well as generalized search and compare our method to several baselines and heuristics. Our evaluation shows that GLS is liked by the users and achieves the shortest interactions.

Learning from a single labeled face and a stream of unlabeled data

Face recognition from a single image per person is a challenging problem because the training sample is extremely small. We study a variation of this problem. In our setting, only a single image of a single person is labeled, and all other people are unlabeled. This setting is very common in authentication on personal computers and mobile devices, and poses an additional challenge because it lacks negative examples. We formalize our problem as one-class classification, and propose and analyze an algorithm that learns a non-parametric model of the face from a single labeled image and a stream of unlabeled data. In many domains, for instance when a person interacts with a computer with a camera, unlabeled data are abundant and easy to utilize. We show how unlabeled data can help in learning better models and evaluate our method on 43 people. The people are identified 90% of the time at nearly zero false positives. This is 15% more often than by Fisherfaces at the same false positive rate. Finally, we conduct a comprehensive sensitivity analysis of our method and provide a guideline for setting its parameters.

Minimal Interaction Content Discovery in Recommender Systems

Many prior works in recommender systems focus on improving the accuracy of item rating predictions. In comparison, the areas of recommendation interfaces and user-recommender interaction remain underexplored. In this work, we look into the interaction of users with the recommendation list, aiming to devise a method that simplifies content discovery and minimizes the cost of reaching an item of interest. We quantify this cost by the number of user interactions (clicks and scrolls) with the recommendation list. To this end, we propose generalized linear search (GLS), an adaptive combination of the established linear and generalized search (GS) approaches. GLS leverages the advantages of these two approaches, and we prove formally that it performs at least as well as GS. We also conduct a thorough experimental evaluation of GLS and compare it to several baselines and heuristic approaches in both an offline and live evaluation. The results of the evaluation show that GLS consistently outperforms the baseline approaches and is also preferred by users. In summary, GLS offers an efficient and easy-to-use means for content discovery in recommender systems.

Does Weather Matter?: Causal Analysis of TV Logs

Weather affects our mood and behaviors, and many aspects of our life. When it is sunny, most people become happier; but when it rains, some people get depressed. Despite this evidence and the abundance of data, weather has mostly been overlooked in the machine learning and data science research. This work presents a causal analysis of how weather affects TV watching patterns. We show that some weather attributes, such as pressure and precipitation, cause major changes in TV watching patterns. To the best of our knowledge, this is the first large-scale causal study of the impact of weather on TV watching patterns.