air transportation


Fast Black-box Variational Inference through Stochastic Trust-Region Optimization

Neural Information Processing Systems

We introduce TrustVI, a fast second-order algorithm for black-box variational inference based on trust-region optimization and the reparameterization trick. At each iteration, TrustVI proposes and assesses a step based on minibatches of draws from the variational distribution. We implemented TrustVI in the Stan framework and compared it to two alternatives: Automatic Differentiation Variational Inference (ADVI) and Hessian-free Stochastic Gradient Variational Inference (HFSGVI). The former is based on stochastic first-order optimization. The latter uses second-order information, but lacks convergence guarantees.


Optimal Black-Box Reductions Between Optimization Objectives

Neural Information Processing Systems

The diverse world of machine learning applications has given rise to a plethora of algorithms and optimization methods, finely tuned to the specific regression or classification task at hand. We reduce the complexity of algorithm design for machine learning by reductions: we develop reductions that take a method developed for one setting and apply it to the entire spectrum of smoothness and strong-convexity in applications. Furthermore, unlike existing results, our new reductions are OPTIMAL and more PRACTICAL. We show how these new reductions give rise to new and faster running times on training linear classifiers for various families of loss functions, and conclude with experiments showing their successes also in practice. Papers published at the Neural Information Processing Systems Conference.


Scalable Inference for Gaussian Process Models with Black-Box Likelihoods

Neural Information Processing Systems

We propose a sparse method for scalable automated variational inference (AVI) in a large class of models with Gaussian process (GP) priors, multiple latent functions, multiple outputs and non-linear likelihoods. Our approach maintains the statistical efficiency property of the original AVI method, requiring only expectations over univariate Gaussian distributions to approximate the posterior with a mixture of Gaussians. Experiments on small datasets for various problems including regression, classification, Log Gaussian Cox processes, and warped GPs show that our method can perform as well as the full method under high levels of sparsity. On larger experiments using the MNIST and the SARCOS datasets we show that our method can provide superior performance to previously published scalable approaches that have been handcrafted to specific likelihood models. Papers published at the Neural Information Processing Systems Conference.


Provable Gradient Variance Guarantees for Black-Box Variational Inference

Neural Information Processing Systems

Recent variational inference methods use stochastic gradient estimators whose variance is not well understood. Theoretical guarantees for these estimators are important to understand when these methods will or will not work. This paper gives bounds for the common "reparameterization" estimators when the target is smooth and the variational family is a location-scale distribution. These bounds are unimprovable and thus provide the best possible guarantees under the stated assumptions. Papers published at the Neural Information Processing Systems Conference.


Explainability: Cracking open the black box, Part 1 - KDnuggets

#artificialintelligence

Explainable AI (XAI) is a sub-field of AI which has been gaining ground in the recent past. And as I machine learning practitioner dealing with customers day in and day out, I can see why. I've been an analytics practitioner for more than 5 years, and I swear, the hardest part of a machine learning project is not creating the perfect model which beats all the benchmarks. It's the part where you convince the customer why and how it works. Humans always had a dichotomy when faced with the unknown.


Aviation is highly engaged with emerging technologies SITA

#artificialintelligence

To someone who works in innovation, it's very reassuring to look at the results in this year's SITA Air Transport IT Insights survey. It shows that airlines and airports are both highly engaged in exploring emerging technologies. Priorities may differ slightly between them, but overridingly it's clear that they're both completely focused on automating and streamlining the journey for passengers. This is an exciting reveal, and it bodes well for the future experience of passengers. The survey shows that when it comes to new technology, the number one focus for investment by airlines continues to be Artificial Intelligence (AI).


Developing a digital twin

#artificialintelligence

In the not too distant future, we can expect to see our skies filled with unmanned aerial vehicles (UAVs) delivering packages, maybe even people, from location to location. In such a world, there will also be a digital twin for each UAV in the fleet: a virtual model that will follow the UAV through its existence, evolving with time. "It's essential that UAVs monitor their structural health," said Karen Willcox, director of the Oden Institute for Computational Engineering and Sciences at The University of Texas at Austin (UT Austin) and an expert in computational aerospace engineering. "And it's essential that they make good decisions that result in good behavior." An invited speaker at the 2019 International Conference for High Performance Computing, Networking, Storage and Analysis (SC19), Willcox shared the details of a project--supported primarily by the U.S. Air Force program in Dynamic Data-Driven Application Systems (DDDAS)--to develop a predictive digital twin for a custom-built UAV.


Artificial Intelligence in Flight Test: A Limited Survey

#artificialintelligence

This year, SFTE's Annual Symposium featured two other papers on very closely related topics. At press time, these authors were not available to share these papers herein.


Chatbots: Basics, Strategy, Best Practices

#artificialintelligence

The idea of using chatbots may seem a little daunting at first. You may be thinking that there's no way your business could ever successfully utilize one, that it's way too complicated, and it won't really help the bottom line of your business -- it's simply not true. Chatbots are not as complicated as you might think. We're here to help give you the information you need to leverage this technology, and we'll also demonstrate how chatbots can absolutely improve your bottom line by automating conversations throughout your business. Below, we give you the basics of chatbots, help you articulate your goals, create a plan with some good old-fashioned chatbot strategy, and share industry best practices.