Despite widespread adoption in NLP, machine learning models remain mostly black boxes. Understanding the reasons behind predictions is, however, quite important in assessing trust in a model. Trust is fundamental if one plans to take action based on a prediction, or when choosing whether or not to deploy a new model. In this work, we describe LIME, a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner. We further present a method to explain models by presenting representative individual predictions and their explanations in a non-redundant manner. We propose a demonstration of these ideas on different NLP tasks such as document classification, politeness detection, and sentiment analysis, with classifiers like neural networks and SVMs. The user interactions include explanations of free-form text, challenging users to identify the better classifier from a pair, and perform basic feature engineering to improve the classifiers.

Tree boosting is a highly effective and widely used machine learning method. In this paper, we describe a scalable endto-end tree boosting system called XGBoost, which is used widely by data scientists to achieve state-of-the-art results on many machine learning challenges. We propose a novel sparsity-aware algorithm for sparse data and weighted quantile sketch for approximate tree learning. More importantly, we provide insights on cache access patterns, data compression and sharding to build a scalable tree boosting system. By combining these insights, XGBoost scales beyond billions of examples using far fewer resources than existing systems.

Given a water distribution network, where should we place sensors to quickly detect contaminants? Or, which blogs should we read to avoid missing important stories? These seemingly different problems share common structure: Outbreak detection can be modeled as selecting nodes (sensor locations, blogs) in a network, in order to detect the spreading of a virus or information as quickly as possible. We present a general methodology for near optimal sensor placement in these and related problems. We demonstrate that many realistic outbreak detection objectives (e.g., detection likelihood, population affected) exhibit the property of "submodularity". We exploit submodularity to develop an efficient algorithm that scales to large problems, achieving near optimal placements, while being 700 times faster than a simple greedy algorithm. We also derive online bounds on the quality of the placements obtained by any algorithm. Our algorithms and bounds also handle cases where nodes (sensor locations, blogs) have different costs. We evaluate our approach on several large real-world problems, including a model of a water distribution network from the EPA, and real blog data. The obtained sensor placements are provably near optimal, providing a constant fraction of the optimal solution. We show that the approach scales, achieving speedups and savings in storage of several orders of magnitude. We also show how the approach leads to deeper insights in both applications, answering multicriteria trade-off, cost-sensitivity and generalization questions.

Despite widespread adoption, machine learning models remain mostly black boxes. Understanding the reasons behind predictions is, however, quite important in assessing trust, which is fundamental if one plans to take action based on a prediction, or when choosing whether to deploy a new model. Such understanding also provides insights into the model, which can be used to transform an untrustworthy model or prediction into a trustworthy one.In this work, we propose LIME, a novel explanation technique that explains the predictions of any classifier in an interpretable and faithful manner, by learning an interpretable model locally around the prediction. We also propose a method to explain models by presenting representative individual predictions and their explanations in a non-redundant way, framing the task as a submodular optimization problem. We demonstrate the flexibility of these methods by explaining different models for text (e.g. random forests) and image classification (e.g. neural networks). We show the utility of explanations via novel experiments, both simulated and with human subjects, on various scenarios that require trust: deciding if one should trust a prediction, choosing between models, improving an untrustworthy classifier, and identifying why a classifier should not be trusted. THE CASE FOR EXPLANATIONSBy "explaining a prediction", we mean presenting textual or visual artifacts that provide qualitative understanding of the relationship between the instance's components (e.g. words in text, patches in an image) and the model's prediction. We argue that explaining predictions is an important aspect in

Declarative queries are proving to be an attractive paradigm for interacting with networks of wireless sensors. The metaphor that "the sensornet is a database" is problematic, however, because sensors do not exhaustively represent the data in the real world. In order to map the raw sensor readings onto physical reality, a model of that reality is required to complement the readings. In this paper, we enrich interactive sensor querying with statistical modeling techniques. We demonstrate that such models can help provide answers that are both more meaningful, and, by introducing approximations with probabilistic confidences, significantly more efficient to compute in both time and energy. Utilizing the combination of a model and live data acquisition raises the challenging optimization problem of selecting the best sensor readings to acquire, balancing the increase in the confidence of our answer against the communication and data acquisition costs in the network. We describe an exponential time algorithm for finding the optimal solution to this optimization problem, and a polynomial-time heuristic for identifying solutions that perform well in practice. We evaluate our approach on several real-world sensor-network data sets, taking into account the real measured data and communication quality, demonstrating that our model-based approach provides a high-fidelity representation of the real phenomena and leads to significant performance gains versus traditional data acquisition techniques.