An article in Wired today is titled “The Genius Neuroscientist Who Might Hold the Key to True AI” .
Professor Karl Friston is a renowned neuroscientist at the Institute of Neurology , University College London (UCL) , the college also incorporates the Leopold Muller Functional Imaging Laboratory (FIL) .
The website for the Wellcome Centre for Human Neuroimaging states the following :
“We bring together clinicians and scientists who study higher cognitive function using neuroimaging techniques. Our goal is to understand how thought and behaviour arise from brain activity, and how such processes break down in neurological and psychiatric disease. Our research groups study all aspects of higher cognitive function including vision, hearing, memory, language, reasoning, emotion, decision making and social interactions.”
All very noble you may think , most scientific research can be used for good or evil . It should also be noted that many scientists are oblivious to the machinations of the controllers , others are fully aware . It should also be noted that large portions of deep state research into the human thought process is often conducted under the guise of ‘understanding mental illness‘ .
Patients with mental health issues make perfect guinea pigs , as do the more unfortunate members of a society .
Karl Friston’s bio-sketch states the folowing :
“Friston currently works on models of functional integration in the human brain and the principles that underlie neuronal interactions. His main contribution to theoretical neurobiology is a free-energy principle for action and perception (active inference).”
Karl Friston studied Natural Sciences (physics and psychology) at the spooky Cambridge University .
Back to the Wired article
The article says that Friston is a ‘heroic figure in academia for devising many of the most important tools that have made human brains legible to science’ . Friston is also credited as inventing statistical parametric mapping , which led to an MRI based neural imaging technique called voxel-based morphometry , itself famous for a study of London taxi drivers .
Friston’s Eureka moment regarding the Free Energy Principle was traced back to the tender age of 8 , when he turned over a log and it appeared that woodlice scattered for the shade . After a while he came to the conclusion that the woodlice were not actually seeking the shade , instead they were simply reacting to the sunlight and as such moved faster .
Friston spent his training years at Littlemore Hospital in Oxfordshire where he worked with schizophrenic patients . To quote Friston on his time there “It was a beautiful place to work ….This little community of intense and florid psychopathology.”
An unusual description for human beings you may think .
After his period at Littlemore , Friston continued in a similar vein with the exception that now he was using PET scans to identify the functioning processes of the human brain .
Shortly after this Friston bumped into Geoffrey Hinton , Geoff studied experimental psychology at the spooky Cambridge University , Geoff also studied artificial intelligence (AI) at Edinburgh University .
According to Wikipedia Geoff is described as the “Godfather of Deep Learning” .
It is also said that Geoff revolutionized the field of computer vision . The field of ‘computer vision’ represents an attempt to emulate human vision .
It is also said that Geoff is one of the founders of a new type of AI system known as capsule neural networks (CapsNet) .
Cultocracy note :
* There will now follow a brief interlude in the article where artificial neural networks will be discussed , please do not fall asleep .
Artificial Neural Network ANN
Convolutional neural networks (CNN) are a type of artificial neural network (ANN) .
In the simple example below you there are three layers ; input , hidden , output .
The circles can be imagined as interconnected neurons in the human brain .
The arrows represent the flow of information , in this case a feedforward network , as the information only flows forward .
Convolutional neural networks (CNN) are a type of ANN often used for visual recognition . They are limited in that they can only identify two dimensional images .
- An Intuitive Explanation of Convolutional Neural Networks
- Convolutional Neural Networks for Visual Recognition
Capsule neural network (CapsNet)
A CapsNet overcomes this issue by nesting neural layers inside other neural layers , forming capsules . As such each ‘capsule’ forming a CapsNet is actually group of ANN’s . Basically ANN’s inside other ANN’s .
Points to note about Geoff Hinton
Also according to Wikipedia :
Hinton moved from the U.S. to Canada in part due to disillusionment with Ronald Reagan-era politics and disapproval of military funding of artificial intelligence. He believes political systems will use AI to “terrorize people”. Hinton has petitioned against lethal autonomous weapons. Regarding existential risk from artificial intelligence, Hinton has stated that superintelligence seems more than 50 years away, but warns that “there is not a good track record of less intelligent things controlling things of greater intelligence”.
Cultocracy note :
Sorry Geoff , AI is already used to terrorize people and worse .
Also Geoff , yes , less intelligent people are in charge , for now at least .
I am sure you realize this .
* End of interlude , you can now wake up if you have been asleep .
The (not so) Great Gatsby
Geoffrey Hinton also co-founded the Gatsby Computational Neuroscience Unit , which is where he met Karl Friston .
By an astonishing coincidence the Gatsby Neuroscience Unit is located next to the FIL .
They were bound to bump into each other at some point .
Cultocracy note :
* There will now follow a brief interlude in the article where corruption & cronyism in UK politics will be discussed , please do not fall asleep .
David (Dave) Sainsbury
According to its official website , Gatsby is a foundation set up by David Sainsbury (of Sainsburys supermarket fame) to realise his ‘charitable objectives’ .
David Sainsbury , or Dave as he prefers to be called , was a major donor to the Labour party under Tony Blair . In fact is has been estimated that Dave gave Tony around £9 million over 5 years , it is also said that Dave donated £2 million each to both Labour & the LibDems in 2001(ish) .
- Ex-Sainsbury’s chair who blew £8m trying to keep us in the EU donates millions to Labour AND the Lib Dems
- New Labour: Donors
Tony’s premiership began in 1997 and ended in 2007 .
Dave was Science Minister in Tony’s government from 1998 until November 2006 .
Dave studied History and Psychology at the spooky Cambridge University .
Dave was also a large donor to the spooky Cambridge University , so large in fact that he was made chancellor of the spooky Cambridge University in 2011 .
It is also said that Dave is pro-EU , I guess Dave didn’t donate enough cash on this occasion .
You can read more about Tony at the link below .
Cultocracy note :
Never shopped at Sainsburys , over priced & pretentious .
No wonder Dave has made so much cash .
* End of interlude , you can now wake up if you have been asleep .
Free Energy Principle
What is the free energy principle ?
Well it appears that nobody really knows , except for Karl Friston that is .
To quote Karl Friston :
“The Free Energy Principle originally emerged from systems neurosciences as a way, a principled way, of understanding what the brain does and how it does it. Subsequently, the principles proved to be so simple and so powerful that they have been applied in a variety of contexts. So one could almost regard the free energy principle as an organizing principle for any living system that shows the characteristics of life.”
At first sight it appears that it is simply an attempt to reverse engineer the human brain , then apply mathematical and computing models to the findings . One of many such efforts .
The Wired article states that “free energy” is a rough synonym for “prediction error” .
Cultocracy note :
Free Energy = Nonconformity
The Bayesian brain hypothesis
Free energy principle relies heavily on Bayesian principles .
In a nutshell and to quote wikipedia : “Bayes’ theorem (alternatively Bayes’ law or Bayes’ rule) describes the probability of an event, based on prior knowledge of conditions that might be related to the event.”
Simple enough , basically the odds of something happening based on prior (a priori) knowledge of similar events . For example , speeding at 200mph on a motorway has led to crashes in 20% of cases , therefore if I speed along a motorway at 200mph then I have a 20% probability of a crash .
What Bayesian based models attempt to do is to narrow the odds based on an as many variables as possible . For example what car am I driving ? Am I drunk ? Is it dark ? Are there other cars on the road ? If so how many ? Is the road wet ? etc….etc…etc…etc…
The more variables that are fed into a Bayesian model , then the greater the chances of predicting a car crash . Theoretically , if you feed in all of the possible variables , then you should be able to predict a crash with a fine degree of accuracy .
Problem , Reaction , Solution
What about the purely human factor , the car crash scenario above relies on a multitude of variables , that is not even accounting for the subset of variables which can actually be applied to the actual driver .
Maybe you could narrow down the variables for a particular human driver and the probability of a crash , for example the drivers age , height , BMI , eyesight , driving experience etc….etc…etc…etc….
Again , the more variables you input the greater the probability of predicting the outcome .
Cultocracy note :
I could simplify the above and narrow it down immediately :
Male (10% chance) or female (90% chance) .
Ahem , only joking .
- Bayes’ theorem (Wikipedia)
- Bayesian probability (Wikipedia)
- An Intuitive Explanation of Bayes’ Theorem (Long Read)
But how do you apply one Bayesian model for a particular person to another person ? Is it like fitting a square peg in a round hole ? Moreover how can you then extrapolate the results so that they will not only fit every person , but all living things ?
Far fetched ? Maybe .
Free energy principle attempts to understand that when a living organism is presented with a specific stimuli , that organism will first react to the stimuli , then the organism will slowly revert back to a neutral state (homeostasis) .
This obviously applies primarily to the neural process .
In simple terms free energy principle attempts to create “a unified brain theory” .
- Free energy principle
- The free-energy principle: a unified brain theory? (PDF)
- The free-energy principle: a rough guide to the brain? (PDF)
So how can we explain ‘free energy’ , not in computing terms , but if applied to a human being ? Free energy could be anything . For example it could be a sensation of pain and the cry of ‘ouch’ when pricked with a needle , it could be a salivous reaction to dinner of steak and wine , it could be a wolf whistle towards a member of the opposite sex (not allowed) , it could be singing and dancing at a celebratory event .
In short the ‘free energy’ that is talked about in free energy theory is everything that makes us human . Of course , we are only talking about free energy in computing terms , aren’t we .
What if you had the capability via computer modelling combined with other technologies to decide what ‘free energy’ to ‘restrict’ in a population ?
In more sinister terms ‘free energy’ could be a deviation from the required discourse , we want red team , he wants blue team , the red team is the neutral state (homeostasis) , the blue team is now seen as ‘free energy’ , which must be reverted to homeostasis .
Even further , what if you could reverse engineer the ‘free energy’ Bayesian computer models and then somehow apply these to actual human beings ? i.e. Reduce a humans ‘free energy’ , regressing them to a neutral state of homeostasis , perhaps permanently .
Reinforcement learning & Ecological Inference
Karl Fristons Free Energy Principle would also appear to be a rival to the currently preferred machine learning method of reinforcement learning , or maybe it is a complimentary approach .
According to DeepMind.com :
“Our goal at DeepMind is to create artificial agents ……..Like a human, our agents learn for themselves to achieve successful strategies that lead to the greatest long-term rewards. This paradigm of learning by trial-and-error, solely from rewards or punishments, is known as reinforcement learning (RL). Also like a human, our agents construct and learn their own knowledge directly from raw inputs, such as vision .”
Karl Fristons talks about active inference , that is maximizing Bayesian model evidence (AI) – and or minimizing variational free energy , which I prefer to call human energy , or at least ‘human type energy’ encountered by an ANN .
Ecological inference “is the process of drawing conclusions about individual-level behavior from aggregate-level data” . Karl Fristons Free Energy Principle would again appear to compliment ecological inference .
Symbolic Convergence Theory
Symbolic convergence theory is a psychological theory .
Wikipedia describes SCT as follows :
“Symbolic Convergence Theory (SCT) is a communication theory developed by Ernest Bormann where people share common fantasies and these collections of individuals are transformed into a cohesive group. SCT offers an explanation for the appearance of a group’s cohesiveness, consisting of shared emotions, motives, and meanings. Through SCT, individuals can build a community or a group consciousness which grows stronger if they share a cluster of fantasy themes.”
What condition could we reach if we combined all of the above machine learning methods , including Karl Fristons free energy principle and also symbolic convergence theory ?
What can we glean from the Wired article about the uses for a ‘free energy’ based AI system ? Well not much , but a few choice quotes may provide a pointer .
“Friston believes he has identified nothing less than the organizing principle of all life, and all intelligence as well.”
“According to the most popular modern Bayesian account, the brain is an “inference engine” that seeks to minimize “prediction error.””
“According to Friston, any biological system9 that resists a tendency to disorder and dissolution will adhere to the free energy principle—whether it’s a protozoan or a pro basketball team.”
Cultocracy note :
What if you can’t resist the ‘tendency to disorder and dissolution’ ?
Got him there !
“And in fact, this is how the free energy principle accounts for everything we do: perception, action, planning, problem solving.”
“To be alive, he says, is to act in ways that reduce the gulf between your expectations and your sensory inputs. Or, in Fristonian terms, it is to minimize free energy.”
“The free energy principle, it turns out, isn’t just a unified theory of action, perception, and planning; it’s also a theory of mental illness.”
“So: The free energy principle offers a unifying explanation for how the mind works and a unifying explanation for how the mind malfunctions. It stands to reason, then, that it might also put us on a path toward building a mind from scratch.”
‘Reinforcement learning doesn’t require humans to label lots of training data; it just requires telling a neural network to seek a certain reward, often victory in a game.’
“On the last day of my trip, I visited Friston in the town of Rickmansworth, where he lives in a house filled with taxidermied animals .”
“As it happens, Rickmansworth appears on the first page of The Hitchhiker’s Guide to the Galaxy; it’s the town where “a girl sitting on her own in a small café” suddenly discovers the secret to making the world “a good and happy place.”
“It’s unclear whether the free energy principle is the secret to making the world a good and happy place, as some of its believers almost seem to think it might be.”
“Friston says his work has two primary motivations. Sure, it would be nice to see the free energy principle lead to true artificial consciousness someday, he says .”
“Rather, his first big desire is to advance schizophrenia research, to help repair the brains of patients like the ones he knew at the old asylum. And his second main motivation, he says, is “much more selfish.” It goes back to that evening in his bedroom, as a teenager, looking at the cherry blossoms, wondering,”
“Can I sort it all out in the simplest way possible?”
Conclusion (Yet again)
AI systems are the future , the technology that exists today is far beyond what is printed in mainstream articles .
Many have had a laugh at Karl Fristons theories , I doubt they would laugh if they knew of the reality .
At the moment AI systems are tailored towards a specific application . I think that Friston is attempting to create a generalized ANN that can be applied towards any person or indeed any living thing , this in turn would serve as a teaching platform for the AI system itself .
Cutting edge AI systems ultimately aim to recreate the creativity and perception of the human mind , combined with the logic and processing power of a computer .
What if you could reverse this by recreating the logic and processing power of a computer , combined with creativity and perception of the human mind ?
As always , the keyword is control .
Further reading :
- The Great Brain Experiment
- The eye in hand: predicting others’ behavior by integrating multiple sources of information (PDF)
- SYMBOLIC CONVERGENCE: A CASE STUDY OF MAGNUM, P.I. (PDF)
- The Active Inference Approach to Ecological Perception: General Information Dynamics for Natural and Artificial Embodied Cognition
- On Monte Carlo methods for Bayesian inference (PDF)
- The Bayesian Approach to Forecasting (PDF)
- Markov blanket
- Markov chain Monte Carlo
- Simulated annealing
- BEYOND BACKPROP AGATION: USING SIMULATED ANNEALING FOR TRAINING NEURAL NETWORKS (PDF)
- Teaching Feed-Forward Neural Networks by Simulated Annealing (PDF)
- SAGRAD: A Program for Neural Network Training with Simulated Annealing and the Conjugate Gradient Method (PDF)
- Designing an autopilot using genetic algorithms and simulated annealing (PDF)
Even More Reading :