Lifeboat Foundation

Google has launched Gemini, a new artificial intelligence system that can seemingly understand and speak intelligently about almost any kind of prompt—pictures, text, speech, music, computer code, and much more.

This type of AI system is known as a multimodal model. It’s a step beyond just being able to handle text or images like previous algorithms. And it provides a strong hint of where AI may be going next: being able to analyze and respond to real-time information from the outside world.

Although Gemini’s capabilities might not be quite as advanced as they seemed in a viral video, which was edited from carefully curated text and still-image prompts, it is clear that AI systems are rapidly advancing. They are heading towards the ability to handle more and more complex inputs and outputs.

Tokamak fusion plasmas benefit from high pressures but are then susceptible to modes of instability. These magnetohydrodynamic (MHD) modes are macroscopic distortions of the plasma, but certain collective motions of individual particles can provide stabilizing effects opposing them. The presence of a resistive wall slows the mode growth, converting a kink to a resistive wall mode (RWM). A kinetic MHD model includes Maxwell’s equations, ideal MHD constraints, and kinetic effects included through the pressure tensor, calculated with the perturbed drift-kinetic distribution function of the particles. The kinetic stabilizing effects on the RWM arise through resonances between the plasma rotation and particle drift motions: precession, bounce, and transit. A match between particle motions and the mode allows efficient transfer of energy that would otherwise drive the growth of the mode, thus damping the growth. The first approach to calculating RWM stability is to write a set of equations for the complex mode frequency in terms of known quantities and then to solve the system. The “energy principle” approach, which has the advantage of clarity in distinguishing the various stabilizing and destabilizing effects, is to change the force balance equation into an equation in terms of changes of kinetic and potential energies, and then to write a dispersion relation for the mode frequency in terms of those quantities. These methods have been used in various benchmarked codes to calculate kinetic effects on RWM stability. The theory has illuminated the important roles of plasma rotation, energetic particles, and collisions in RWM stability.

Research conducted by a team of scientists from Kaunas universities, Lithuania, revealed that low-frequency ultrasound influences blood parameters. The findings suggest that ultrasound’s effect on haemoglobin can improve oxygen’s transfer from the lungs to bodily tissues.

The research was undertaken on 300 blood samples collected from 42 pulmonary patients. The samples were exposed to six different low-frequency ultrasound modes at the Institute of Mechatronics of Kaunas University of Technology (KTU).

The changes in 20 blood parameters were registered using the blood analysing equipment at the Lithuanian University of Health Sciences (LSMU) laboratories. For the prediction of ultrasound exposure, artificial intelligence, i.e. analysis of variance (ANOVA), non-parametric Kruskal-Wallis method and machine learning algorithms were applied. The calculations were made at the KTU Artificial Intelligence Centre.

Mastercard has announced that it has developed an in-house generative AI to help combat fraud on its payment processing network.

Instead of relying on textual inputs, Mastercard’s algorithm uses a cardholder’s merchant visit history as a prompt to determine whether a transaction involves a business that the customer would likely visit. The algorithm generates pathways through Mastercard’s network, akin to heat-sensing radar, to provide a score as an answer.

Continue reading “Mastercard’s new anti-fraud AI has a success rate of up to 300%” »

An AI algorithm outperformed other screening methods in identifying cervical precancer. The approach could be especially valuable in low-resource settings.

There has been significant progress in the field of quantum computing. Big global players, such as Google and IBM, are already offering cloud-based quantum computing services. However, quantum computers cannot yet help with problems that occur when standard computers reach the limits of their capacities because the availability of qubits or quantum bits, i.e., the basic units of quantum information, is still insufficient.

One of the reasons for this is that bare qubits are not of immediate use for running a quantum algorithm. While the binary bits of customary computers store information in the form of fixed values of either 0 or 1, qubits can represent 0 and 1 at one and the same time, bringing probability as to their value into play. This is known as quantum superposition.

This makes them very susceptible to external influences, which means that the information they store can readily be lost. In order to ensure that quantum computers supply reliable results, it is necessary to generate a genuine entanglement to join together several physical qubits to form a logical qubit. Should one of these physical qubits fail, the other qubits will retain the information. However, one of the main difficulties preventing the development of functional quantum computers is the large number of physical qubits required.

Window to the soul? Maybe, but the eyes are also a flashing neon sign for a new artificial intelligence-based system that can read them to predict what you’ll do next.

A University of Maryland researcher and two colleagues have used eye-tracking technology and a new deep-learning AI algorithm to predict study participants’ choices while they viewed a comparison website with rows and columns of products and their features.

The algorithm, known as RETINA (Raw Eye Tracking and Image Ncoder Architecture), could accurately zero in on selections before people had even made their decisions.

Researchers mapped over 10,000 mouse hippocampal #neurons, creating the world’s most comprehensive database of single-neuron #connectivity #patterns.

Summary: Researchers unveiled the most extensive single-neuron projectome database to date, featuring over 10,000 mouse hippocampal neurons.

The study provides an unprecedented view of the spatial connectivity patterns at the mesoscopic level, crucial for understanding learning, memory, and emotional processing in the hippocampus. By employing machine learning algorithms for categorizing axonal trajectories and integrating spatial transcriptome data, researchers identified 43 distinct projectome cell types, revealing intricate projection patterns and soma locations’ correspondence to projection targets.

Continue reading “Mapping the Brain: The Largest Neuron Projectome Unveiled” »

Why is Adam the most popular optimizer in Deep Learning? Let’s understand it by diving into its math, and recreating the algorithm.