Understanding the predictions of a machine learning model can be as crucial as the model's accuracy in many application domains. However, the black-box nature of most highly-accurate (complex) models is a major hindrance to their interpretability. To address this issue, we introduce the symbolic metamodeling framework -- a general methodology for interpreting predictions by converting "black-box" models into "white-box" functions that are understandable to human subjects. A symbolic metamodel is a model of a model, i.e., a surrogate model of a trained (machine learning) model expressed through a succinct symbolic expression that comprises familiar mathematical functions and can be subjected to symbolic manipulation. We parameterize symbolic metamodels using Meijer G-functions -- a class of complex-valued contour integrals that depend on scalar parameters, and whose solutions reduce to familiar elementary, algebraic, analytic and closed-form functions for different parameter settings.
Transfer learning algorithms are used when one has sufficient training data for one supervised learning task (the source/training domain) but only very limited training data for a second task (the target/test domain) that is similar but not identical to the first. Previous work on transfer learning has focused on relatively restricted settings, where specific parts of the model are considered to be carried over between tasks. Recent work on covariate shift focuses on matching the marginal distributions on observations $X$ across domains. Similarly, work on target/conditional shift focuses on matching marginal distributions on labels $Y$ and adjusting conditional distributions $P(X Y)$, such that $P(X)$ can be matched across domains. However, covariate shift assumes that the support of test $P(X)$ is contained in the support of training $P(X)$, i.e., the training set is richer than the test set.
We study the problem of black-box optimization of a function $f$ of any dimension, given function evaluations perturbed by noise. The function is assumed to be locally smooth around one of its global optima, but this smoothness is unknown. Our contribution is an adaptive optimization algorithm, POO or parallel optimistic optimization, that is able to deal with this setting. POO performs almost as well as the best known algorithms requiring the knowledge of the smoothness. Furthermore, POO works for a larger class of functions than what was previously considered, especially for functions that are difficult to optimize, in a very precise sense.
A multitude of factors can contribute to a flight being delayed, but Air France, who partnered with a handful of other companies, is testing the world's first self-driving luggage tug in hopes of streamlining airport operations and improving the speed of getting luggage to and from an aircraft. The vehicle, known as the AT135 baggage tractor, began official testing at France's Toulouse-Blagnac airport last month on November 15. To the untrained eye it looks like the myriad of vehicles you already see scurrying around the airport tarmac while waiting for a flight, including a cab with a seat, steering wheel, and all the controls needed for a human driver. But look closer and you'll be able to spot some of the telltale hardware upgrades of an autonomous vehicle, including laser scanning LIDAR sensors on the roof and bumper that complement less visible sensors like GPS and front and rear cameras providing a 360-degree view around the tug. Climb inside the tug and you'll also find a big switch allowing it to be switched between manual and autonomous modes, as well as an oversized touchscreen showing a map of the airport and all the gates the vehicle is designed to service.
The widespread adoption of autonomous vehicles (AVs) seems inevitable. Despite various concerns, AVs' development and implementation continues to advance. How will their spread affect sustainability? How will they affect humans' capacity to live on Earth in a way that does not threaten the planet's life-support function? We asked experts the following question: "How will the proliferation of autonomous vehicles affect sustainability?"
Do you know the way to San Jose? As they previewed earlier this year, Bosch and Mercedes-Benz have commenced trials for an automated ride-hailing service in the Silicon Valley city of San Jose. To start with, autonomous S-Class Mercedes-Benz vehicles (with safety drivers at the wheel) will shuttle "a select group of users" between North San Jose and downtown. The busy San Carlos/Stevens Creek corridor between west San Jose and downtown should be good test for the self-driving tech used by Mercedes and Bosch. Rather than just playing with prototypes, the companies want to create a production-ready SAE Level 4/5 self-driving system that can be built into different makes and models.
Colin Parris has a challenging job. As the vice president of software and analytics research at General Electric, Parris must evaluate new technology and applications that can benefit the manufacturing giant. All of which must work within the framework that GE employs when assessing safety and efficiency known as Humble AI. But even after his own rigorous evaluation and approval process there is no easy way to get buy-in from the rest of the company for new ways of doing business. With huge investments in aviation systems, energy and healthcare, GE is always looking for ways to use technology to improve operations, deliver products faster and better anticipate problems along the way -- all areas where AI can potentially be of use.
Me: "Alexa, tell me what will happen in 2020." Amazon AI: "Here's what I found on Wikipedia: The 2020 UEFA European Football Championship…[continues to read from Wikipedia]" Me: "Alexa, give me a prediction for 2020." Amazon AI: "The universe has not revealed the answer to me." Well, some slight improvement over last year's responses, when Alexa's answer to the first question was "Do you want to open'this day in history'?" As for the universe, it is an open book for the 120 senior executives featured here, all involved with AI, delivering 2020 predictions for a wide range of topics: Autonomous vehicles, deepfakes, small data, voice and natural language processing, human and augmented intelligence, bias and explainability, edge and IoT processing, and many promising applications of artificial intelligence and machine learning technologies and tools. And there will be even more 2020 AI predictions, in a second installment to be posted here later this month. "Vehicle AI is going to be ...
The mobility revolution may have the potential to transform cities, but in the short term the rise in ride-hailing apps, bike sharing, and electric scooters is giving many local officials fits. A healthy dose of data and machine learning may help get this movement back on track. That's the bet that San Francisco-based StreetLight Data is making. The company is helping cities harness the explosion of data being generated by everything from smart city sensors to mobile phones to new transportation modes, in a bid to reinvent urban planning. As cities groan under rising populations and pollution, making more effective use of data could be the key to making them habitable over the long run.
Robots and artificial intelligence haven't taken over -- yet. But machines still have human-like needs, needs that, when fulfilled, can improve payments and commerce and, according to some, usher in what's sometimes being called in smart circles a new industrial revolution. The key to that future is emerging 5G mobile network technology. Sure, retailers, healthcare providers and other operations with firm footing in the digital economy are seeking, finding and crafting use cases for 5G, which, above all, promises speedier and more robust information flows among people and businesses. But machine-to-machine communication promises to play as great a role or even greater in 5G, at least according to some observers and studies.