Lifeboat Foundation

About five years ago, Areg Danagoulian, associate professor in the MIT Department of Nuclear Science and Engineering (NSE), became intrigued by a technique developed by researchers at Los Alamos National Laboratory that uses a neutron beam to identify unknown materials.

“They could look into a black box containing uranium and say what kind and how much,” says Danagoulian, who directs MIT’s Laboratory of Applied Nuclear Physics (LANPh). “I was thinking about the problem of verifying nuclear material in warheads, and it just dawned on me, this amazing technology could be applied to what we’re working on.”

But there was a problem: This method, called neutron resonance transmission analysis (NRTA), requires an enormous, expensive apparatus, limiting its utility for the kind of on-site nuclear material applications Danagoulian and his research colleagues focus on. To leapfrog this obstacle, they determined to make NRTA technology portable.

Glow is an iconic interesting research about deep neural networks that can generalize with small training sets.

Since the early days of machine learning, artificial intelligence scenarios have faced with two big challenges in order to experience mainstream adoption. First, we have the data efficiency problem that requires machine or deep learning models to be trained using large and accurate datasets which, as we know, are really expensive to build and maintain. Secondly, we have the generalization problem which AI agents face in order to build new knowledge that is different from the training data. Humans, by contrast, are incredibly efficient learning with minimum supervision and rapidly generalizing knowledge from a few data examples.

Generative models are one of the deep learning disciplines that focuses on addressing the two challenges mentioned above. Conceptually, generative models are focused on observing an initial dataset, like a set of pictures, and try to learn how the data was generated. Using more mathematical terms, generative models try to infer all dependencies within very high-dimensional input data, usually specified in the form of a full joint probability distribution. Entire deep learning areas such as speech synthesis or semi-supervised learning are based on generative models. Recently, generative models such as generative adversarial networks(GANs) have become extremely popular within the deep learning community. Recently, OpenAI experimented with a not-very well-known technique called Flow-Based Generative Models in order to improve over existing methods.

Continue reading “This Neural Networkfrom OpenAI can Learn from Small Datasets” »

Sorry, we’re having trouble playing this video.

Learn More

Contains 24 logical fallacies, 24 cognitive biases, 3 Game Cards and 3 Call Out Cards. Perfect for honing critical thinking skills, calling out people you’re arguing with, and spotting actual fake news.

Continue reading “Critical Thinking Cards Deck” »

At the heart of almost every sufficiently massive galaxy there is a black hole whose gravitational field, although very intense, affects only a small region around the center of the galaxy. Even though these objects are thousands of millions of times smaller than their host galaxies, our current view is that the Universe can be understood only if the evolution of galaxies is regulated by the activity of these black holes, because without them the observed properties of the galaxies cannot be explained.

Theoretical predictions suggest that as these black holes grow they generate sufficient energy to heat up and drive out the gas within galaxies to great distances. Observing and describing the mechanism by which this energy interacts with galaxies and modifies their evolution is therefore a basic question in present day Astrophysics.

With this aim in mind, a study led by Ignacio Martín Navarro, a researcher at the Instituto de Astrofísica de Canarias (IAC), has gone a step further and has tried to see whether the matter and energy emitted from around these black holes can alter the evolution, not only of the host galaxy, but also of the satellite galaxies around it, at even greater distances. To do this, the team has used the Sloan Digital Sky Survey, which allowed them to analyze the properties of the galaxies in thousands of groups and clusters. The conclusions of this study, started during Navarro’s stay at the Max Planck Institute for Astrophysics, are published today in Nature magazine.

In your business you need to learn how to distinguish between chatbots and conversational AI, here are some tips on how to do that.

In recent years, roboticists have developed a wide variety of robots with human-like capabilities. This includes robots with bodies that structurally resemble those of humans, also known as humanoid robots.

Testing the performance of humanoid robots can sometimes be challenging, as there are numerous measures to consider when trying to determine their applicability in real-world scenarios. Two features that are particularly important for humanoid robots are posture control and balance, as these robot’s body structures can sometimes make them prone to falling or stumbling, especially in complex environments.

Researchers at Technische Universität Berlin and the University Clinic of Freiburg recently created a system to evaluate the posture control and balance of both humans and humanoid robots. This system, presented in a paper pre-published on arXiv, is designed to assess balance and posture control of robots or humans as they perform different movements on a moving surface.

China’s going all in on deep learning. The Beijing Academy of Artificial Intelligence (BAAI) recently released details concerning its “Wu Dao” AI system – and there’s a lot to unpack here.

Up front: Wu Dao is a multi-modal AI system. That means it can do a bunch of different things. It can generate text, audio, and images, and, according to Engadget, it can even “power virtual idols.”

The reason for all the hullabaloo surrounding Wu Dao involves its size. This AI model is huge. It was trained using a whopping 1.75 trillion parameters. For comparison, OpenAI’s biggest model, GPT-3, was trained with just 175 billion.

Google has helped create the most detailed map yet of the connections within the human brain. It reveals a staggering amount of detail, including patterns of connections between neurons, as well as what may be a new kind of neuron.

The brain map, which is freely available online, includes 50000 cells, all rendered in three dimensions. They are joined together by hundreds of millions of spidery tendrils, forming 130 million connections called synapses. The data set measures 1.4 petabytes, roughly 700 times the storage capacity of an average modern computer.

The data set is so large that the researchers haven’t studied it in detail, says Viren Jain at Google Research in Mountain View, California. He compares it to the human genome, which is still being explored 20 years after the first drafts were published.

‘I would expect these vehicles to make 15 trips, twice a day during rush hour to justify the cost of the vehicles,’ says researcher.

The key is to rapidly heat the battery to a certain temperature using a nickel foil, which then allows for ultra quick charging without causing any damage.

“I think flying cars have the potential to eliminate a lot of time and increase productivity and open the sky corridors to transportation,” Dr Wang said.

Continue reading “Battery breakthrough makes flying cars commercially viable” »