Randomized experiments are the gold standard for evaluating the effects of changes to real-world systems, including Internet services. Data in these tests may be difficult to collect and outcomes may have high variance, resulting in potentially large measurement error. Bayesian optimization is a promising technique for optimizing multiple continuous parameters for field experiments, but existing approaches degrade in performance when the noise level is high. We derive an exact expression for expected improvement under greedy batch optimization with noisy observations and noisy constraints, and develop a quasi-Monte Carlo approximation that allows it to be efficiently optimized. Experiments with synthetic functions show that optimization performance on noisy, constrained problems outperforms existing methods. We further demonstrate the effectiveness of the method with two real experiments conducted at Facebook: optimizing a production ranking system, and optimizing web server compiler flags.

An ensemble of models (EM), where each model is constructed on a diverse subset of feature variables, is proposed to rank rare class items ahead of majority class items in a highly unbalanced two class problem. The proposed ensemble relies on an algorithm to group the feature variables into subsets where the variables in a subset work better together in a model and the variables in different subsets work better in separate models. The strength of the EM depends on the algorithm's ability to identify strong and diverse subsets of feature variables. A second phase of ensembling is achieved by aggregating several EMs each optimized on a diverse evaluation metric. The resulting ensemble is called ensemble of models and metrics (EMM). Here, the diverse/complementary evaluation metrics ensure increased diversity among EMs to aggregate. The ensembles are applied to the protein homology data, downloaded from the 2004 KDD cup competition website, to rank proteins in such a way that the rare homologous proteins are found ahead of the majority non-homologous proteins. The ensembles are constructed using feature variables which are various scores from sequence alignments of amino acids in a candidate protein and three dimensional descriptions of a native protein representing functional and structural similarity of proteins. While prediction performances of the EMs are better than the contemporary state-of-the-art ensembles and competitive to the winning procedures of the $2004$ KDD cup competition, the performances of the EMM are found on the top of all. In this application, we have two diverse EMs constructed on two complementary evaluation metrics average precision and rank last, where the former is robust against ranking close homologs and the latter is robust against ranking distant homologs. The advantage of using EMM is that it is robust against both close and distant homologs.

Exploiting the wealth of imaging and non-imaging information for disease prediction tasks requires models capable of representing, at the same time, individual features as well as data associations between subjects from potentially large populations. Graphs provide a natural framework for such tasks, yet previous graph-based approaches focus on pairwise similarities without modelling the subjects' individual characteristics and features. On the other hand, relying solely on subject-specific imaging feature vectors fails to model the interaction and similarity between subjects, which can reduce performance. In this paper, we introduce the novel concept of Graph Convolutional Networks (GCN) for brain analysis in populations, combining imaging and non-imaging data. We represent populations as a sparse graph where its vertices are associated with image-based feature vectors and the edges encode phenotypic information. This structure was used to train a GCN model on partially labelled graphs, aiming to infer the classes of unlabelled nodes from the node features and pairwise associations between subjects. We demonstrate the potential of the method on the challenging ADNI and ABIDE databases, as a proof of concept of the benefit from integrating contextual information in classification tasks. This has a clear impact on the quality of the predictions, leading to 69.5% accuracy for ABIDE (outperforming the current state of the art of 66.8%) and 77% for ADNI for prediction of MCI conversion, significantly outperforming standard linear classifiers where only individual features are considered.

Human visual object recognition is typically rapid and seemingly effortless, as well as largely independent of viewpoint and object orientation. Until very recently, animate visual systems were the only ones capable of this remarkable computational feat. This has changed with the rise of a class of computer vision algorithms called deep neural networks (DNNs) that achieve human-level classification performance on object recognition tasks. Furthermore, a growing number of studies report similarities in the way DNNs and the human visual system process objects, suggesting that current DNNs may be good models of human visual object recognition. Yet there clearly exist important architectural and processing differences between state-of-the-art DNNs and the primate visual system. The potential behavioural consequences of these differences are not well understood. We aim to address this issue by comparing human and DNN generalisation abilities towards image degradations. We find the human visual system to be more robust to image manipulations like contrast reduction, additive noise or novel eidolon-distortions. In addition, we find progressively diverging classification error-patterns between man and DNNs when the signal gets weaker, indicating that there may still be marked differences in the way humans and current DNNs perform visual object recognition. We envision that our findings as well as our carefully measured and freely available behavioural datasets provide a new useful benchmark for the computer vision community to improve the robustness of DNNs and a motivation for neuroscientists to search for mechanisms in the brain that could facilitate this robustness.

Extracting per-frame features using convolutional neural networks for real-time processing of video data is currently mainly performed on powerful GPU-accelerated workstations and compute clusters. However, there are many applications such as smart surveillance cameras that require or would benefit from on-site processing. To this end, we propose and evaluate a novel algorithm for change-based evaluation of CNNs for video data recorded with a static camera setting, exploiting the spatio-temporal sparsity of pixel changes. We achieve an average speed-up of 8.6x over a cuDNN baseline on a realistic benchmark with a negligible accuracy loss of less than 0.1% and no retraining of the network. The resulting energy efficiency is 10x higher than that of per-frame evaluation and reaches an equivalent of 328 GOp/s/W on the Tegra X1 platform.

The conversation in the technology community about artificial intelligence was first rekindled by manufacturing giant IBM and its AI platform, Watson, CEO Ginni Rometty said on Tuesday. "We are the ones that woke up the AI world here again," Rometty told "Mad Money" host Jim Cramer in a wide-ranging interview about Washington, Warren Buffett and her business. Rometty said that the key to her century-old company remaining an institution in this country is how many times it has been able to reinvent itself and follow the latest trends in tech. Today, those trends are the cloud and artificial intelligence, which IBM employees refer to as "cognitive" programming. "There's a reason we call it cognitive," Rometty told Cramer.

When applied to previously-collected atmospheric samples and data, their findings support evidence that on average these bioaerosols globally make up less than 1 percent of the particles in the upper troposphere -- where they could influence cloud formation and by extension, the climate -- and not around 25 to 50 percent as some previous research suggests. While atmospheric and climate modeling suggests that bioaerosols, globally averaged, are not abundant and efficient enough at freezing to significantly influence cloud formation, research findings have varied significantly. The group leveraged the presence of phosphorus in the mass spectra to train the classification machine learning algorithm on known samples and then, primed, applied it to field data acquired from Desert Research Institute's Storm Peak Laboratory in Steamboat Springs, Colorado, and from the Carbonaceous Aerosol and Radiative Effects Study based in the town of Cool, California. Knowing that the principal atmospheric emissions of phosphorus are from mineral dust, combustion products, and biological particles, they exploited the presence of phosphate and organic nitrogen ions and their characteristic ratios in known samples to classify the particles.

This is a class in data visualization. But before we leap into making charts and maps, we'll consider the nature of data, and some basic principles that will help you to "interview" datasets to find and tell stories. This is not a class in statistics, but I will introduce a few fundamental statistical concepts, which hopefully will stand you in good stead as we work to visualize data over the next few weeks -- and beyond. We're often told that there are "lies, damned lies, and statistics." But data visualization and statistics provide a view of the world that we can't otherwise obtain. They give us a framework to make sense of daunting and otherwise meaningless masses of information. The "lies" that data and graphics can tell arise when people misuse statistics and visualization methods, not when they are used correctly. The best data journalists understand that statistics and graphics go hand-in-hand. Just as numbers can be made to lie, graphics may misinform if the designer is ignorant of or abuses basic statistical principles. You don't have to be an expert statistician to make effective charts and maps, but understanding some basic principles will help you to tell a convincing and compelling story -- enlightening rather than misleading your audience. I hope you will get hooked on the power of a statistical way of thinking. As data artist Martin Wattenberg of Google has said: "Visualization is a gateway drug to statistics." Download the data for this session from here, unzip the folder and place it on your desktop.

I often get pitched with a superior deep learning solution for Natural Language Understanding (NLU). After all, deep learning is the disruptive new force in AI. A better NLU AI entices many useful advancements, ranging from smarter chat bots and virtual assistants to news categorization, with an ultimate promise of better language comprehension. Lets assume this superior deep learning (DL) "product" is called "(dot)AI". Their pitch deck will invariably have a bar chart that looks something like this -- the claim being that the new DL topic classifier/tagger of (Dot)AI is better than state of the art methods.

A man killed in a crash last year while using the semi-autonomous driving system on his Tesla Model S sedan kept his hands off the wheel for extended periods of time despite repeated automated warnings not to do so, a US government report said on Monday. The National Transportation Safety Board released 500 pages of findings into the May 2016 death of Joshua Brown, a former Navy SEAL, near Williston, Florida. Brown's Model S collided with a truck while it was engaged in the'Autopilot' mode and he was killed. A Tesla Inc spokeswoman Tesla spokeswoman Keely Sulprizio declined to comment on the NTSB report. Lawyers for Brown's family did not return messages seeking comment.