The encoder network of an autoencoder is an approximation of the nearest point projection onto the manifold spanned by the decoder. A concern with this approximation is that, while the output of the encoder is always unique, the projection can possibly have infinitely many values. This implies that the latent representations learned by the autoencoder can be misleading. Borrowing from geometric measure theory, we introduce the idea of using the reach of the manifold spanned by the decoder to determine if an optimal encoder exists for a given dataset and decoder. We develop a local generalization of this reach and propose a numerical estimator thereof. We demonstrate that this allows us to determine which observations can be expected to have a unique, and thereby trustworthy, latent representation. As our local reach estimator is differentiable, we investigate its usage as a regularizer and show that this leads to learned manifolds for which projections are more often unique than without regularization.

In this article, we get to learn how industrial robots have been a dominant force in the manufacturing industry for decades. The amount of robots being installed has skyrocketed he recent years and it will continue to do so until 2021. This is an interesting read for anyone that wants to learn more about the robotic industry! The global industrial robot installation reaches a peak in 2021, with half a million units put to work and new robots put in the market. One of the biggest news stories recently has been the increase in industrial robots used across the world.

As intractable problems accrue and grow, artificial intelligence is increasingly being called upon as part of the solution. Carnegie Mellon University AI researchers have stepped up to help surmount these obstacles where large data sets must be analyzed and patterns discovered to find answers. Last year, the National Science Foundation teamed up with the U.S. Department of Agriculture, the U.S. Department of Homeland Security, as well as corporate sponsors Accenture, Amazon, Google and Intel to provide $220 million in grants to create 11 new institutes specifically dedicated to AI research across a wide range of sectors. CMU's School of Computer Science and College of Engineering faculty will work with four of these new institutes: the AI Institute for Resilient Agriculture, the AI Institute for Collaborative Assistance and Responsive Interaction for Networked Groups, the AI Institute for Future Edge Networks and Distributed Intelligence, and the USDA-NIFA Institute for Agricultural AI for Transforming the Workforce and Decision Support. Learn more about these institutes and meet the researchers leading the work in our magazine, The Link.

The expansive role that artificial intelligence (AI) can play is being valued on a global stage for its transformative merits, in plenty of industries, to improve efficiencies and outcomes. But there are also ethical questions around the responsible use of AI for a globalized society, governed by a respect for human rights and democratic principles. Powerful AI-driven technologies have the potential to be misused, and there have already been numerous studies on AI displaying bias towards certain skin shades, amplifying discrimination against women, and members of marginalized ethnic groups. Minority groups often become the unwitting victim of ethical AI conundrums, in large part because machine learning works best off large subsets of data, with the majority of input data coming from, well, the majority. While these AI-driven discrepancies are well-recognized, the power of AI as a transformative force for societal good is also coming to be well-regarded, with research from the Pew Center indicating that over half (53%) of people from 20 countries surveyed saying that artificial intelligence has had a positive effect on society.

As customer and employee bases become increasingly global, many organizations are finding the value of enterprise video is negatively affected. The fact is, the inability to quickly and seamlessly serve video assets in multiple languages significantly lowers the value of video as a communication media. But thanks to advances in AI technology, organizations of any size can now add captions to video assets--8 times faster than was previously possible. During this Webcast the presenters will cover the concept and technology behind captioning, as well as how leading brands are using it as a core element in their video production workflow. The presenters will also cover how AI plays a part in the process and how search engine results are affected--and give a live demonstration of the latest transcription and translation technology.