Lifeboat Foundation

Artificial intelligence (AI) has the potential to transform technologies as diverse as solar panels, in-body medical sensors and self-driving vehicles. But these applications are already pushing today’s computers to their limits when it comes to speed, memory size and energy use.

Fortunately, scientists in the fields of AI, computing and nanoscience are working to overcome these challenges, and they are using their brains as their models.

That is because the circuits, or neurons, in the human brain have a key advantage over today’s computer circuits: they can store information and process it in the same place. This makes them exceptionally fast and energy efficient. That is why scientists are now exploring how to use materials measured in billionths of a meter— nanomaterials—to construct circuits that work like our neurons. To do so successfully, however, scientists must understand precisely what is happening within these nanomaterial circuits at the atomic level.

Such as complex models of various areas of the cortex. The main novelty of this work is the abstraction of a neuromorphic architecture into clusters represented by minicolumns and hypercolumns, analogously to the fundamental structural units observed in neurobiology. Without this approach, simulating large-scale fully connected networks needs prohibitively large memory to store look-up tables for point-to-point connections. Instead, we use a novel architecture, based on the structural connectivity in the neocortex, such that all the required parameters and connections can be stored in on-chip memory. The cortex simulator can be easily reconfigured for simulating different neural networks without any change in hardware structure by programming the memory. A hierarchical communication scheme allows one neuron to have a fan-out of up to 200 k neurons. As a proof-of-concept, an implementation on one Altera Stratix V FPGA was able to simulate 20 million to 2.6 billion leaky-integrate-and-fire (LIF) neurons in real time. We verified the system by emulating a simplified auditory cortex (with 100 million neurons). This cortex simulator achieved a low power dissipation of 1.62 μW per neuron. With the advent of commercially available FPGA boards, our system offers an accessible and scalable tool for the design, real-time simulation, and analysis of large-scale spiking neural networks.

Our inability to simulate neural networks in software on a scale comparable to the human brain (10¹¹ neurons, 10¹⁴ synapses) is impeding our progress toward understanding the signal processing in large networks in the brain and toward building applications based on that understanding. A small-scale linear approximation of a large spiking neural network will not be capable of providing sufficient information about the global behavior of such highly nonlinear networks. Hence, in addition to smaller scale systems with detailed software or hardware neural models, it is necessary to develop a hardware architecture that is capable of simulating neural networks comparable to the human brain in terms of scale, with models with an intermediate level of biological detail, that can simulate these networks quickly, preferably in real time to allow interaction between the simulation and the environment.

Artificial Intelligence (AI) and Deep Learning, with a focus on Natural Language Processing (NLP), have seen substantial changes in the last few years. The area has advanced quickly in both theoretical development and practical applications, from the early days of Recurrent Neural Networks (RNNs) to the current dominance of Transformer models.

Models that are capable of processing and producing natural language with efficiency have advanced significantly as a result of research and development in the field of neural networks, particularly with regard to managing sequences. RNN’s innate ability to process sequential data makes them well-suited for tasks involving sequences, such as time-series data, text, and speech. Though RNNs are ideally suited for these kinds of jobs, there are still problems with scalability and training complexity, particularly with lengthy sequences.

https://marktechpost-newsletter.beehiiv.com/subscribe.

One of the cornerstone challenges in machine learning, time series forecasting has made groundbreaking contributions to several domains. However, forecasting models can’t generalize the distribution shift that changes with time because time series data is inherently non-stationary. Based on the assumptions about the inter-instance and intra-instance temporal distribution shifts, two main types of techniques have been suggested to address this issue. Both stationary and nonstationary dependencies can be separated using these techniques. Existing approaches help reduce the impact of the shift in the temporal distribution. Still, they are overly prescriptive because, without known environmental labels, every sequence instance or segment might not be stable.

Before learning about the changes in the stationary and nonstationary states throughout time, there is a need to identify when the shift in the temporal distribution takes place. By assuming nonstationarity in observations, it is possible to theoretically identify the latent environments and stationary/nonstationary variables according to this understanding.

https://marktechpost-newsletter.beehiiv.com/subscribe.

Generative AI can be hijacked by self-replicating worms.

The great mystery of where all the aliens are in our vast Universe contemplates ancient interstellar civilizations building enormous megastructures that rival worlds or even stars in the immensity… and asks why we can’t see these giant alien artifacts.

David Brin on Event Horizon with John Michael Godier: • A.I. Wars, The Fermi Paradox and Grea…
This Week in Space with Rod Pyle: • Alien Megastructures — Isaac Arthur a…

Continue reading “The Fermi Paradox: Absent Megastructures” »

Johns Hopkins electrical and computer engineers are pioneering a new approach to creating neural network chips—neuromorphic accelerators that could power energy-efficient, real-time machine intelligence for next-generation embodied systems like autonomous vehicles and robots.

Electrical and computer engineering graduate student Michael Tomlinson and undergraduate Joe Li—both members of the Andreou Lab—used natural language prompts and ChatGPT4 to produce detailed instructions to build a spiking neural network chip: one that operates much like the human brain.

Through step-by-step prompts to ChatGPT4, starting with mimicking a single biological neuron and then linking more to form a network, they generated a full chip design that could be fabricated.

[email protected] AI: Viewing Artificial Intelligence Through the Lens of the Theory of Everything Kindle EditionDive into Conscious AI: Vi…

Gödel’s Incompleteness theorems are two theorems of mathematical logic that demonstrate the inherent limitations of every formal axiomatic system capable of modelling basic arithmetic.

The first incompleteness theorem: No consistent formal system capable of modelling basic arithmetic can be used to prove all truths about arithmetic.

In other words, no matter how complex a system of mathematics is, there will always be some statements about numbers that cannot be proved or disproved within the system.