Google has caved to pressure from its staff and abandoned a new AI ethics panel after hundreds demanded conservative members of the board were sacked for their views. The search giant announced last week that it was setting up a new board to tackle moral issues surrounding its use of the technology. It hoped to avoid controversies by using a broad spectrum of expertise to inform its future decisions, but the move has ironically stirred up a debacle of its own. Eight experts from outside the company were recruited and employees at the traditionally liberal leaning firm took issue with two of the appointees. More than 1,000 of its protest-prone workers signed an open letter objecting to specific board members, who they say are'anti-trans' and pro-military drones.

This paper develops methodology for 3D radial visualization of high-dimensional datasets. Our display engine is called RadViz3D and extends the classic RadViz that visualizes multivariate data in the 2D plane by mapping every record to a point inside the unit circle. The classic RadViz display has equally-spaced anchor points on the unit circle, with each of them associated with an attribute or feature of the dataset. RadViz3D obtains equi-spaced anchor points exactly for the five Platonic solids and approximately for the other cases via a Fibonacci grid. We show that distributing anchor points at least approximately uniformly on the 3D unit sphere provides a better visualization than in 2D. We also propose a Max-Ratio Projection (MRP) method that utilizes the group information in high dimensions to provide distinctive lower-dimensional projections that are then displayed using Radviz3D. Our methodology is extended to datasets with discrete and mixed features where a generalized distributional transform is used in conjuction with copula models before applying MRP and RadViz3D visualization.

Data-to-text generation can be conceptually divided into two parts: ordering and structuring the information (planning), and generating fluent language describing the information (realization). Modern neural generation systems conflate these two steps into a single end-to-end differentiable system. We propose to split the generation process into a symbolic text-planning stage that is faithful to the input, followed by a neural generation stage that focuses only on realization. For training a plan-to-text generator, we present a method for matching reference texts to their corresponding text plans. For inference time, we describe a method for selecting high-quality text plans for new inputs. We implement and evaluate our approach on the WebNLG benchmark. Our results demonstrate that decoupling text planning from neural realization indeed improves the system's reliability and adequacy while maintaining fluent output. We observe improvements both in BLEU scores and in manual evaluations. Another benefit of our approach is the ability to output diverse realizations of the same input, paving the way to explicit control over the generated text structure.

Estimating conditional dependence graphs and precision matrices are some of the most common problems in modern statistics and machine learning. When data are fully observed, penalized maximum likelihood-type estimators have become standard tools for estimating graphical models under sparsity conditions. Extensions of these methods to more complex settings where data are contaminated with additive or multiplicative noise have been developed in recent years. In these settings, however, the relative performance of different methods is not well understood and algorithmic gaps still exist. In particular, in high-dimensional settings these methods require using non-positive semidefinite matrices as inputs, presenting novel optimization challenges. We develop an alternating direction method of multipliers (ADMM) algorithm for these problems, providing a feasible algorithm to estimate precision matrices with indefinite input and potentially nonconvex penalties. We compare this method with existing alternative solutions and empirically characterize the tradeoffs between them. Finally, we use this method to explore the networks among US senators estimated from voting records data.

In this paper, we describe our submission to SemEval-2019 Task 4 on Hyperpartisan News Detection. Our system relies on a variety of engineered features originally used to detect propaganda. This is based on the assumption that biased messages are propagandistic in the sense that they promote a particular political cause or viewpoint. We trained a logistic regression model with features ranging from simple bag-of-words to vocabulary richness and text readability features. Our system achieved 72.9% accuracy on the test data that is annotated manually and 60.8% on the test data that is annotated with distant supervision. Additional experiments showed that significant performance improvements can be achieved with better feature pre-processing.

Google is ending a new artificial intelligence ethics council just one week after launching it, following protests from employees over the appointment of a rightwing thinktank leader. The rapid downfall of the Advanced Technology External Advisory Council (ATEAC), which was dedicated to "the responsible development of AI", came after more than 2,000 Google workers signed a petition criticizing the company's selection of an anti-LGBT advocate. "It's become clear that in the current environment, ATEAC can't function as we wanted. So we're ending the council and going back to the drawing board," a Google spokesperson told the Guardian in a statement on Thursday. Google faced intense backlash soon after announcing that one of the eight council members was Kay Coles James, the president of the Heritage Foundation, a conservative thinktank with close ties to Donald Trump's administration.

WASHINGTON - Embattled U.S. aviation giant Boeing on Thursday insisted on the "fundamental safety" of its 737 Max aircraft but pledged to take all necessary steps to ensure the jets' airworthiness. The statements came hours after Ethiopian officials said pilots of a doomed plane that crashed last month, leaving 157 people dead, had followed the company's recommendations. The preliminary findings released Thursday by transportation authorities in Addis Ababa put the American aircraft giant under even greater pressure to restore public trust amid mounting signs the company's onboard anti-stall systems were at fault in crashes involving its formerly top-selling 737 Max aircraft -- incidents that left nearly 350 people dead in less than five months. "We remain confident in the fundamental safety of the 737 Max," CEO Dennis Muilenburg said in a statement, adding that impending software fixes would make the aircraft "among the safest airplanes ever to fly." Muilenburg also acknowledged, however, that an "erroneous activation" of Boeing's Maneuvering Characteristics Augmentation System had occurred. The system is designed to prevent stalls but may have forced the Ethiopian and Indonesian jets into the ground.

Google's Advanced Technology External Advisory Council was supposed to oversee its work on artificial intelligence and ensure it doesn't cross any lines. Now, the council wouldn't be able to do any of that, because the tech giant has officially cancelled it just a bit over a week after it was announced. According to Vox, the project was falling apart from the start due to Google's decision to name controversial figures as members of the board. The most problematic of them was perhaps Kay Coles James, the president of Heritage Foundation, which has long advocated against LGBT rights. The group also has a long history of climate change denial and anti-immigrant sentiments.

In an earlier post today, we distilled half a dozen DARPA-dense docs into an easy-to-follow overview of the DARPA Subterranean Challenge (SubT), a new competition that will task teams of humans and robots to explore complex underground environments. In this post, we have an interview with SubT program manager Timothy Chung, whom we met late last year at DARPA's D60 Conference. "I think for many of the technologies we're seeking to advance--it's one of those, aim for the moon, even if you miss you hit the stars type of an approach," he told us about the new challenge. "So we envision some component technologies being immediately operationally of value, but we've set the bar ambitiously high enough for it to be DARPA-worthy and also provide a vision for how that kind of impact could be magnified if and when we're successful." IEEE Spectrum: What are the SubT courses going to be like?

As part of the very first event in the DARPA Subterranean Challenge (SubT), the organizers have invited nine teams (and their robots) to Edgar Experimental Mine in Idaho Springs, Colo., for a sort of test run called the SubT Integration Exercise, or STIX. These nine teams have already demonstrated their systems to DARPA, showing that they can navigate autonomously over rough terrain, locate objects, and respond to an e-stop command if they go berserk. For the teams, this will be an opportunity to test out their robots in an actual tunnel system, and at the same time DARPA itself will be able to make sure all of their testing infrastructure and whatnot works, well in advance of the Tunnel Circuit Challenge itself, which will take place in August. Our detailed post on SubT and interview with DARPA program manager Timothy Chung cover all of this stuff, along with the guidelines that teams have to follow when designing and deploying their systems, but all that information doesn't necessarily give a sense of what kind of hardware teams will likely be deploying at SubT. Fortunately, many of the teams participating in STIX have posted pictures or videos of their robots, so we've put together this article to introduce each team and have a look at what they'll be working with.