Anybody else getting permission errors at
git submodule update --init --recursive ?
Another route which might be an alternative route to intelligent computation, or at least some interesting areas. Note the mention of ants.
Photonic elementary cellular automata for simulation of complex phenomena
Abstract
Cellular automata are a class of computational models based on simple rules and algorithms that can simulate a wide range of complex phenomena. However, when using conventional computers, these ‘simple’ rules are only encapsulated at the level of software. This can be taken one step further by simplifying the underlying physical hardware. Here, we propose and implement a simple photonic hardware platform for simulating complex phenomena based on cellular automata. Using this special-purpose computer, we experimentally demonstrate complex phenomena, including fractals, chaos, and solitons, which are typically associated with much more complex physical systems. The flexibility and programmability of our photonic computer present new opportunities to simulate and harness complexity for efficient, robust, and decentralized information processing using light.
Introduction
Modern digital electronic computers, which are based on the von Neumann architecture, exhibit extreme hardware complexity in their construction and are composed of billions of transistors engineered in a hierarchical and highly structured manner. Unlike the von Neumann architecture, nature is abound with emergent phenomena and complex systems containing many interacting components following simple rules with no hierarchical control. For example, social insects like ants with only limited local information can collectively self-organize to form global structures1. This suggests that an alternative and potentially more efficient way to simulate such phenomena is to harness simple and decentralized physical hardware that directly emulates the underlying rules of a complex system.
One class of computational models that can benefit from simple and decentralized physical hardware is cellular automata (CA), which exhibits complex behavior emerging from the local interactions of cells arranged on a regular lattice2. CA were introduced in the 1940s to study how self-replication and evolution can emerge in artificial life3 and was later popularized in Conway’s Game of Life4, which exhibits self-organizing patterns reminiscent of biological systems. Subsequent landmark studies revealed that CA are also capable of replicating other complex behavior such as fractals5, chaos6, self-organized criticality7, synchronization8, and universal computation9. Consequently, CA have found utility in modeling a wide range of natural phenomena in physics10,11, chemistry12,13,14, and biology15. Furthermore, CA have important applications in real-world computational problems such as cryptography16, data compression17, error-correction18, simulating traffic flow19, and developing more robust artificial intelligence20. Owing to their simple formulations, certain CA of interest are computationally irreducible21, i.e., there are no analytical shortcuts to evaluate their state after an arbitrary time without resorting to executing the sequential simulation in its entirety. On the other hand, most CA are only implemented as high-level software on conventional computers, resulting in unnecessary overhead. Therefore, it is desirable to seek out physical hardware that better encapsulates the computational principles of CA to enable more efficient simulation. Notable previous attempts to implement physical systems tailored to perform CA include self-assembling DNA molecules22, arrays of nanomagnets23, memristor networks24, and living slime molds25.
In this work, we propose and experimentally implement a photonic computational platform capable of simulating complex phenomena using CA. Compared to other approaches, our photonic platform offers several distinct advantages: (1) the inherently high bandwidth endowed by computing using light offers potentially orders-of-magnitude speed-up in clock rate over the simulation of CA on conventional von Neumann computers, (2) rapid reconfigurability for easy programming of a variety of CA rules enables many different complex phenomena to be observed in the same physical system, and (3) the kind of sparse, local, and shift-invariant connections required for CA are well-suited for this platform. We will demonstrate how even simple photonic hardware can host a wide range of complex emergent phenomena and is capable of sophisticated (or even universal) computation. By exploiting this complexity, we reveal a path toward the next generation of more efficient or robust photonic hardware accelerators for reservoir computing26,27 and deep learning20,28.
6 Likes
Genetic programming and CA are truly interesting. I think we cannot underestimate them at all. They can show emergent behaviour very quickly. I really was heavily into macroevolution (mentally at least).
The llm models though are showing something very strange and unexpected wrt intelligence, however w want to measure that.
The whole thing is a big bowl of flux right now.
A more recent thing that is even more dramatic in my eyes is liquid neural networks, i.e. single layer networks that not only encode/decode like transformers, but learn dynamically and do so in a way we can almost understand. Just a few equations for the nodes/neurons and they seem to show dynamic and real time learning. LLM // transformers cannot (yet) do that.
9 Likes
Neat to hear about the advancements and implementing in hardware. I’ve been intrigued by CA ever since reading Wolfram’s “A new kind of science” circa 2003.
2 Likes
Tapping in a computer password while chatting over Zoom could open the door to a cyber-attack, research suggests, after a study revealed artificial intelligence (AI) can work out which keys are being pressed by eavesdropping on the sound of the typing.
Experts say that as video conferencing tools such as Zoom have grown in use, and devices with built-in microphones have become ubiquitous, the threat of cyber-attacks based on sounds has also risen.
4 Likes
Everyone able to code… 
:
1 Like
They are going to be adding a GPT webcrawler
3 Likes
3 Likes
Initially, CoreWeave was focused exclusively on cryptocurrency applications. But it pivoted within the last several years to general-purpose computing as well as generative AI technologies, like text-generating AI models.
Fast-forward to today and CoreWeave provides access to over a dozen SKUs of Nvidia GPUs in the cloud, including H100s, A100s, A40s and RTX A6000s, for use cases like AI and machine learning, visual effects and rendering, batch processing and pixel streaming.
“Our clients include generative AI companies, like Tarteel AI and Anlatan, the creators of NovelAI, and we’ve supported a range of open source AI and machine learning projects like EleutherAI and Stability AI’s Stable Diffusion,” Intrator told TechCrunch in an email interview. “We also work with a number of notable VFX and animation studios such as Spire Animation, and partner closely with 3D streaming and ‘metaverse’ companies such as PureWeb.”
It’s tough for any cloud provider to compete with the incumbents in the space — i.e., Google, Amazon and Microsoft. For perspective, AWS made $80.1 billion in revenue last year, while Google Cloud and Azure made $75.3 billion and $26.28 billion, respectively.
Safe Network with compute will make this look like a huge waste. Even without compute! lol
5 Likes
I dunno. A large amount of highly specialized compute might be a ways off for SAFE … but if this company were to offer it’s services on SAFE for SNT, then perhaps it’s win-win.
2 Likes
3 Likes
It all started to go wrong when they upped the wokeness? 
Thats a bit OTT, gotta be honest 
but unintended consequences…
Its still a very new field and a vast amount to be learned yet but the last line of the article is worth repeating
“It’s just mimicking reasoning, rather than actually performing that reasoning.”
4 Likes
A lot of humanity does the same thing though.
2 Likes
That is the closest thing to self awareness I’ve seen, I take that as a breakthrough. No wonder there is flooding outside, it’s difficult to take in that such rare things can happen. keep up the good work.
This is an extension on the keypad to enter code to open door. Security doors used to be able to be cracked by looking at the marks left by fingers on the glass. Including phone unlock swipes.
still I wonder what success they have from only listening to the sound.
1 Like
90% and respected to increase.
2 Likes
A counter to that (and it exists) is a keypad (with LED display or so) where the position of the keys changes each time.
And concerning that sound method: it doesn’t help with copy paste of passwords from a password manager, for example. I also read something about playing random key touches as counter measure, but who is going to do that.
1 Like
More than 40 million voters may have had their details stolen in the biggest data breach in British history.
The Electoral Commission revealed yesterday that “hostile actors” had access to its systems for 14 months without detection, and that hackers may have obtained the name and address of almost every voter in the UK who registered to vote between 2014 and 2022.
Russian hackers are the main suspects in the intrusion.
I’m curious to know if the UK Electoral Commission (and beyond) will be interested in transferring data once SafeNet is implemented.
This helps out those pushing a Digital ID.
2 Likes
