Lifeboat Foundation

After breaking all the records related to training computer vision models, NVIDIA now claims that it’s AI platform is able to train a natural language neural network model based on one of the largest datasets in a record time. It also claims that the inference time is just 2 milliseconds which translates to an extremely fast response from the model participating in a conversation with a user.

After computer vision, natural language processing is one of the top applications of AI. From Siri to Alexa to Cortana to Google Assistant, all conversational user experiences are powered by AI.

The advancements in AI research is putting the power of language understanding and conversational interface into the hands of developers. Data scientists and developers can now build custom AI models that work exactly like Alexa and Siri but for a specialized and highly customized industry use case from the healthcare or legal vertical. This enables doctors and lawyers to interact with expert agents that can understand the terminology and the context of the conversation. This new user experience is going to be a part of future line of business applications.

Harvard University researchers have developed a new powered exosuit that can make you feel as much as a dozen pounds lighter when walking or running. Scientific American reports that the 11-pound system, which is built around a pair of flexible shorts and a motor worn on the lower back, could benefit anyone who has to cover large distances by foot, including recreational hikers, military personnel, and rescue workers.

According to the researchers, who have published their findings in the journal Science, this system differs from previous exosuits because it’s able to make it easier to both walk and run. The challenge, as shown by a video accompanying the research, is that your legs work very differently depending on whether you’re walking or running. When walking, the team says your center of mass moves like an “inverted pendulum,” while running causes it to move like a “spring-mass system.” The system needs to be able to accommodate both of them, and sense when the wearer’s gait changes.

One day, who knows when, artificial intelligence could hollow out the job market. But for now, it is generating relatively low-paying jobs. The market for data labeling passed $500 million in 2018 and it will reach $1.2 billion by 2023, according to the research firm Cognilytica. This kind of work, the study showed, accounted for 80% of the time spent building AI technology.

Is the work exploitative? It depends on where you live and what you’re working on. In India, it is a ticket to the middle class. In New Orleans, it’s a decent enough job. For someone working as an independent contractor, it is often a dead end.

There are skills that must be learned — like spotting signs of a disease in a video or medical scan or keeping a steady hand when drawing a digital lasso around the image of a car or a tree. In some cases, when the task involves medical videos, pornography or violent images, the work turns grisly.

This is already happening on a small testing scale; the Big roll out is coming in 4 or 5 years.

UPS has been delivering a new kind of automated mail — and it’s not via email.

Continue reading “UPS Has Been Secretly Delivering Mail in Self-Driving Trucks for Months” »

Twenty years ago, entertainment was dominated by a handful of producers and monolithic broadcasters, a near-impossible market to break into.

And now, over 50 years later, AI is bringing stories to life like we’ve never seen before.

Converging with the rise of virtual reality and colossal virtual worlds, AI has begun to create vastly detailed renderings of dead stars, generate complex supporting characters with intricate story arcs, and even bring your favorite stars—whether Marlon Brando or Amy Winehouse—back to the big screen and into a built environment.

Continue reading “AI Is About to Completely Change the Face of Entertainment” »

A trio of physicists from the National Autonomous University of Mexico and Tec de Monterrey has solved a 2,000-year-old optical problem—the Wasserman-Wolf problem. In their paper published in the journal Applied Optics, Rafael González-Acuña, Héctor Chaparro-Romo, and Julio Gutiérrez-Vega outline the math involved in solving the puzzle, give some examples of possible applications, and describe the efficiency of the results when tested.

Over 2,000 years ago, Greek scientist Diocles recognized a problem with optical lenses—when looking through devices equipped with them, the edges appeared fuzzier than the center. In his writings, he proposed that the effect occurs because the lenses were spherical—light striking at an angle could not be focused because of differences in refraction. Isaac Newton was reportedly stumped in his efforts to solve the problem (which became known as spherical aberration), as was Gottfried Leibniz.

In 1949, Wasserman and Wolf devised an analytical means for describing the problem, and gave it an official name—the Wasserman-Wolf problem. They suggested that the best approach to solving the problem would be to use two aspheric adjacent surfaces to correct aberrations. Since that time, researchers and engineers have come up with a variety of ways to fix the problem in specific applications—most particularly cameras and telescopes. Most such efforts have involved creating aspherical lenses to counteract refraction problems. And while they have resulted in improvement, the solutions have generally been expensive and inadequate for some applications.

Technology that translates cortical activity into speech would be transformative for people unable to communicate as a result of neurological impairment. Decoding speech from neural activity is challenging because speaking requires extremely precise and dynamic control of multiple vocal tract articulators on the order of milliseconds. Here, we designed a neural decoder that explicitly leverages the continuous kinematic and sound representations encoded in cortical activity to generate fluent and intelligible speech. A recurrent neural network first decoded direct cortical recordings into vocal tract movement representations, and then transformed those representations to acoustic speech output. Modeling the articulatory dynamics of speech significantly enhanced performance with limited data. Naïve listeners were able to accurately identify and transcribe decoded sentences. Additionally, speech decoding was not only effective for audibly produced speech, but also when participants silently mimed speech. These results advance the development of speech neuroprosthetic technology to restore spoken communication in patients with disabling neurological disorders.

Robots are about to go underground — for a competition anyways.

The Defense Advanced Research Projects Agency (DARPA), the branch of the U.S. Department of Defense dedicated to developing new emerging technologies, is holding a challenge intended to develop technology for first responders and the military to map, navigate, and search underground. But the technology developed for the competition could also be used in future NASA missions to caves and lava tubes on other planets.

Continue reading “NASA Robots to Compete in Underground Challenge in Mining Tunnels” »

This presentation was posted by Jason Mayes, senior creative engineer at Google, and was shared by many data scientists on social networks. Chances are that you might have seen it already. Below are a few of the slides. The presentation provides a list of machine learning algorithms and applications, in very simple words. It also explain the differences between AI, ML and DL (deep learning.)