GPT: Diagnosing Alzheimers through Speech

Posted by Peter Rudin on 20. January 2023 in News No Comments

Research in Neuroscience suggests that indicators of Alzheimer’s disease could be detected early: a subtle change in speech patterns, grammatical mistakes, forgetting the meaning of a word, or mispronouncing common words that used to flow naturally.

A team from Drexel University took a major step in applying GPT-3’s capacity to detect text-patterns, using a massive dataset of interviews that included patients with and without Alzheimers. Based on this approach, information needed to extract speech patterns that could be applied to identify markers of Alzheimers in future patients is now available.

“This could be very useful for early screening and risk assessment before a clinical diagnosis”, one of the authors of the study suggests.

Read More

‘Liquid’ Machine-Learning with New Algorithms

Posted by Peter Rudin on 6. January 2023 in News No Comments

Last year MIT researchers announced that they had built ‘liquid’ neural networks, inspired by the brains of small species. The flexibility of these ‘liquid’ neural nets yielded better decision-making for many tasks.

But these models became computationally expensive as their number of neurons and synapses needed to be increased to solve their underlying complicated math.

Now, the same team of scientists has discovered a way to alleviate this bottleneck with a new type of fast and efficient artificial intelligence algorithms.

These new network models have the same characteristics of ‘liquid’ neural nets – flexible, causal, robust, and explainable –  but they are orders of magnitude faster, and scalable.

Read More

Silent Night

Posted by Peter Rudin on 23. December 2022 in News No Comments

A difficult year is coming to an end. Humanity is faced with a cluster of unprecedented difficulties, ranging from climate issues to war and the struggle for political power. Our democracy and privacy are at risk.

Meanwhile Christmas is celebrated in a tradition of joy and reunion. Gifts are exchanged as a sign of one’s gratitude. The tune of  ‘Silent Night’, a Christmas carol composed in 1818 by Franz Xaver Gruber, is heard throughout the night, touching our emotions.

‘Silent Night’ also offers the opportunity to slow down and reflect on one’s inner peace. Following this thought, I have added the link to an Essay I wrote over two years ago which – in my view – still holds true. With that done, I wish you a very Merry Christmas!

Read More

Intel Unveils Real-Time Deepfake Detector

Posted by Peter Rudin on 9. December 2022 in News No Comments

Intel just introduced ‘FakeCatcher’, a real-time detector of deepfakes with a claimed  accuracy rate of 96%. It is focused on clues within actual videos, measuring the amount of light that is absorbed or reflected by facial blood vessels.

Deepfake detection has become increasingly important as interactive deepfakes offer the illusion of talking to a real person and compositional deepfakes create many deepfakes to compile a ‘synthetic history’ around an individual’s personality.

We are still far away to determine whether a piece of text, audio or video is human-generated or not. Intel’s deepfake detector could present a significant step forward since its detection method offers a new AI-supported approach to uncover video-fakes.

Read More

IBM Reaches Qubit-Count Over 400 With New Processor

Posted by Peter Rudin on 25. November 2022 in News No Comments

IBM just announced the latest generation of quantum processors with more than three times the qubit-count of previous models. Next the company plans  to release the first 1,000-qubit processors in 2023.

If qubit-counts are sufficiently high and error rates low enough, it will be possible that re-running specific calculations several times, will provide answers to problems that are difficult or impossible to achieve with conventional computers.

As a major effort that is  part of  its roadmap, IBM is working on  the development of new software. The goal is to support developers to implement algorithms for quantum computers without having to worry about issuing all the discrete control commands that perform the actual manipulations of qubits.

Read More

Predictions 2023: AI Will Become a Coworker

Posted by Peter Rudin on 11. November 2022 in News No Comments

Forrester Research’s just released predictions report for AI, highlights what most have already observed: AI adoption has evolved from a nice-to-have trend to experiment with, to a must-do priority.

In 2023 AI adoption will “continue to expand and be more creative, trustworthy and optimized.” This will happen across almost all industry sectors.

AI-powered virtual assistants will engage with customers and employees. At the same time human-produced content creation will never be fast enough to address the need for personalized content.

Hence, many companies plan to invest in AI-supported digital content creation, for example to provide quick and easy summarization of complex textual topics.

Read More

Intel’s new Neuromorphic Approach to Robotics

Posted by Peter Rudin on 28. October 2022 in News No Comments

Neuromorphic computing replicates the neural structure of the human brain to create algorithms that deal with the uncertainties of the natural world with far less energy compared to conventional computing.

Intel has developed one of the most notable architectures in the field: the Loihi neuromorphic chip.
Intel’s Lab tested the approach in a simulated 3D environment. In this setup, the robot actively senses objects by moving an event-based camera that functions as its eyes.

We now need to test the algorithm in the real-world with actual robots .“Our goal is to apply similar capabilities to future robots that work in interactive settings, enabling them to adapt to the unforeseen and work more naturally alongside humans”.

Read More

Virtual Reality(VR) and the Limits of the Human Eye

Posted by Peter Rudin on 14. October 2022 in News No Comments

Silicon Valley is eager to get us excited about the coming of the metaverse, but today’s virtual reality hardware is a long way from meeting their ambitious goals. 

One of the biggest challenges is building better displays with far more pixels per inch, but experts say new materials and designs are on the way.

Researchers from Samsung and Stanford University believe that emerging technologies could soon get us close to the theoretical limit of pixel density.

While the required improvements in eye and head tracking add extra complexity to designs, the authors say this is probably the innovation that  will happen soonest.

Read More

Decoding Speech From Thoughts Without Brain Surgery

Posted by Peter Rudin on 30. September 2022 in News No Comments

The advanced pattern-recognition capabilities of AI deep learning offer new glimmer of hope of a non-invasive solution to help patients who are unable to speak due to neurodegenerative diseases.

To address this challenge, Meta AI researchers used AI deep learning, specifically a convolutional neural network (CNN), to help decode brain activity using data captured without open brain surgery.

As for the societal impact, the Meta AI researcher caution, “Although these results hold great promise for the development of a safe and scalable system to help patients with communication deficits, the scientific community should remain vigilant that it will not be adapted to decode brain signals without the consent of the participants.”

Read More

Yann LeCun’s Vision for the Future of Machine Learning

Posted by Peter Rudin on 16. September 2022 in News No Comments

Professor Yann LeCun, CDS faculty & founding Director and Chief AI Scientist at Meta’s AI lab, wants machines to operate with common sense.

He has built an early version of his world model that is capable of basic object recognition and is currently working on teaching it to make predictions.

LeCun’s new approach is based on a neural network that would be able to process the world at various levels of detail. This network would focus only on those features that are relevant for the task at hand.

His vision places the world model and the configurator as two key elements in a larger system or cognitive architecture that would include other neural networks. 

Read More