Lifeboat Foundation: Safeguarding Humanity
Toggle light / dark theme

Blog

Category: robotics/AI – Page 1,434

First AI laser skin treatment

A tech company in Hong Kong says it has developed the world’s first artificially intelligent laser skin treatment, which scans and detects the heat, sensitivity and shape of a customer’s face. Rods Technology spent over five years developing the technology to help reduce the number of injuries caused by manual treatments conducted by “blind” dermatologists. Over 100 laser facial injuries were reported in Hong Kong between January and July 1, 2022. With a human operator only needing to turn on the machine, the robot dermatologist is able to customise a treatment to help reduce acne scarring, wrinkles and even remove tattoos. The first commercial facial was sold in July 2022.

comment.

The role of AI in the People’s Liberation Army

The People’s Liberation Army (PLA) of China is likely one of the leading forces in AI development as far as investment is concerned. An October 2021 report published by the Centre for Security and Emerging Technology (CSET) at Georgetown University estimated that the PLA was spending between $1.6bn and $2.7bn on AI research and procurement per year, which is approximately equivalent to that of the US military.

The report, titled Harnessed Lightning, identified seven areas of interest for the PLA and its AI development that are detailed here in order of the quantity of contracts awarded as found by CSET:

It is notable that the priority area for the PLA is the development of autonomous vehicles, specifically sub-surface and aerial platforms. This suggests that the primary concern at present is the development of autonomous platforms that would be able to contribute to generating an asymmetric advantage for the PLA in combat with the US or a similarly advanced opponent.

Ant Colonies Behave Like Neural Networks When Making Decisions

Summary: Researchers suggest that when in a group, ants behave in a similar fashion to networks of neurons in the brain.

Source: Rockefeller University.

Temperatures are rising, and one colony of ants will soon have to make a collective decision. Each ant feels the rising heat beneath its feet but carries along as usual until, suddenly, the ants reverse course. The whole group rushes out as one—a decision to evacuate has been made. It is almost as if the colony of ants has a greater, collective mind.

Is a neuron’s information processing more complex than a perceptron?

I’ve been researching the relationship between brain neurons and nodes in neural networks. Repeatedly it is claimed neurons can do complex information processing that vastly exceeds that of a simple activation function in a neural network.

The resources I’ve read so far suggest nothing fancy is happening with a neuron. The neuron sums the incoming signals from synapses, and then fires when the sum passes a threshold. This is identical to the simple perceptron, the precursor to today’s fancy neural networks. If there is more to a neuron’s operation that this, I am missing it due to lack of familiarity with the neuroscience terminology. I’ve also perused this stack exchange, and haven’t found anything.

If someone could point to a detailed resource that explains the different complex ways a neuron processes the incoming information, in particular what makes a neuron a more sophisticated information processor than a perceptron, I would be grateful.

Deep learning methods for designing proteins scaffolding functional sites

Current approaches to de novo design of proteins harboring a desired binding or catalytic motif require pre-specification of an overall fold or secondary structure composition, and hence considerable trial and error can be required to identify protein structures capable of scaffolding an arbitrary functional site. Here we describe two complementary approaches to the general functional site design problem that employ the RosettaFold and AlphaFold neural networks which map input sequences to predicted structures. In the first “constrained hallucination” approach, we carry out gradient descent in sequence space to optimize a loss function which simultaneously rewards recapitulation of the desired functional site and the ideality of the surrounding scaffold, supplemented with problem-specific interaction terms, to design candidate immunogens presenting epitopes recognized by neutralizing antibodies, receptor traps for escape-resistant viral inhibition, metalloproteins and enzymes, and target binding proteins with designed interfaces expanding around known binding motifs. In the second “missing information recovery” approach, we start from the desired functional site and jointly fill in the missing sequence and structure information needed to complete the protein in a single forward pass through an updated RoseTTAFold trained to recover sequence from structure in addition to structure from sequence. We show that the two approaches have considerable synergy, and AlphaFold2 structure prediction calculations suggest that the approaches can accurately generate proteins containing a very wide array of functional sites.

The authors have declared no competing interest.