Articles tagged with Neural Networks
A Nagoya University research team discovered two distinct thermosensory pathways in the brain that transmit temperature information to different areas of the forebrain. Blocking one pathway was found to impair a rat's ability to avoid heat, while blocking another impaired its ability to avoid cold.
New research shows that electric fields drive neural activity and coordinate memory across two key brain regions. The findings could lead to improved brain-controlled prosthetics for people with paralysis.
The study reveals five distinct brain types, each suited for its purpose, from a jellyfish's diffuse neural network to the human brain's reflective capabilities. Researchers suggest that autonomous machines can learn from coordination in bees, rapid thinking in birds, and single-mindedness in worms.
Researchers at EPFL have found a way to teach quantum computers to learn and process information using principles inspired by quantum mechanics. By training quantum neural networks (QNNs) on a few simple examples called 'product states', the computer can effectively grasp complex dynamics of entangled quantum systems.
Researchers use photonics to accelerate convolutional neural networks by harnessing light's unique properties, reducing power consumption and increasing efficiency. This approach enables real-time image processing and scalable solutions for complex images.
A team of researchers from Kyoto University and international institutions has developed a mathematical solution to the temporal asymmetry of nonequilibrium disordered Ising networks. This breakthrough offers insights into the behavior of biological systems, machine learning, and AI tools.
Researchers develop an ionic device utilizing redox reactions to achieve a high number of reservoir states, enabling efficient complex nonlinear operations. The device demonstrated remarkable performance in solving second-order nonlinear dynamic equations and predicting future values with low mean square prediction error.
Researchers found that neural networks grown in a dish can predict future stimulus events, with higher prediction efficiencies in focally stimulated networks. Focal electrical stimulation induced long-term memory traces and reduced dependence on short-term memory.
Scientists at Karolinska Institutet have identified a group of nerve cells involved in creating negative emotional states and chronic stress. The neurons, which are sensitive to oestrogen levels, were mapped using advanced techniques such as Patch-seq, Neuropixels, and optogenetics.
BioAutoMATED is an all-in-one AutoML platform designed for biologists, enabling easy analysis and interpretation of biological sequences. The platform uses three existing AutoML tools to generate models that can predict biological functions from sequence information.
The proposed multi-scale facial video pulse extraction network uses separable spatiotemporal convolution and dimension separable attention to extract pulse signals from facial videos. This approach yields better results than single-scale methods, especially when fusing multiple scales.
The study reveals that spontaneous waves of neurotransmitter glutamate facilitate dendrite pruning, while a unique protection/punishment machinery strengthens certain connections and eliminates others. Proper pruning is critical for neural development, with insufficient or excessive connections linked to neurophysiological disorders.
Researchers at Mainz University and Berlin found a previously unknown function of electrical synapses in the insect neural network, governing wing movement and generating consistent flight power. The discovery reveals new concepts of information processing by the central nervous system.
Scientists using popular computational tools to interpret AI predictions are picking up too much 'noise' when analyzing DNA. Researchers have found a way to fix this by applying a new line of code, leading to more reliable explanations and potentially unlocking the next breakthrough in health and medicine.
Researchers trained a robotic chef to watch and learn from cooking videos, enabling it to identify ingredients and actions. The robot recognized 93% of the correct recipe from 16 videos, including variations and new recipes, showcasing its potential for automated food production and cost-effective deployment.
A unique microcircuit in fruit flies' visual system transforms a single type of neuronal input to compute direction selectivity, with no inhibitory neurons present. The discovery reveals a striking example of the multilayered mechanisms of inhibition and excitation in the brain.
A deep learning-based framework called EMGSense enables accurate wearable EMG device usage through AI self-training techniques. It achieves average accuracy of 91.9% in gesture recognition and 81.2% in activity recognition, outperforming state-of-the-art approaches.
A study by Florida Atlantic University and CINVESTAV found that long-term running maintains the connectivity of adult-born hippocampal neurons, which contribute to memory function during aging. Exercise may prevent or delay age-related memory decline by increasing the survival and modifying the network of these neurons.
Research suggests that brain's electrical activity tunes sub-cellular components to optimize network stability and efficiency. This 'Cytoelectric Coupling' hypothesis proposes that electric fields influence neurons' physical configuration to fine-tune information processing.
Researchers developed a new sensor array that mimics the red, green and blue photoreceptors in human eyes, producing high-quality images through a neural network-based algorithm. The device has the potential to revolutionize camera technology by increasing spatial resolution and reducing power consumption.
Glioblastoma steals cognitive faculties as it spreads, but its insidious ability to infiltrate neighboring networks may be its undoing. Researchers found neural activity can restructure connections in surrounding tissue, causing decline. The drug gabapentin blocks this growth-causing activity in mice with glioblastoma.
A deep neural network developed by researchers at the University of California - Santa Cruz has been shown to accurately classify particle signals with 99.8% accuracy in real-time. The system can identify weak or noisy signals and pinpoint their source, making it suitable for point-of-care applications.
Researchers have developed photonic neural networks that can achieve precision comparable to conventional neural networks but with considerable energy savings. The devices use a programmable grid of silicon interferometers to perform calculations in under 0.1 nanoseconds, paving the way for faster and more efficient AI applications.
Researchers developed GAME-Net, a graph neural network that rapidly evaluates adsorption energy for large molecules like plastics and biomass. The model achieves accuracy comparable to density functional theory (DFT) while utilizing simple molecular representations.
A deep learning model has been developed to classify cancer cells into distinct types, enabling accurate prediction of metastatic potential. The tool achieves high accuracy and is simple to use, making it a promising solution for medical practitioners.
Researchers at Sainsbury Wellcome Centre found that instinctual exploratory runs enable mice to learn a map of the world efficiently. The study demonstrates how biological brains can learn faster and more efficiently than AI agents by focusing on salient objects.
A team of researchers developed a self-checking deep learning system that accurately extracts information from gravitational-wave data. The algorithm, called DINGO, has been trained to interpret real data and can cross-check its own results for accuracy.
Researchers developed a new neural network, MSUN, that accurately classifies plant diseases in natural settings using transfer learning and unsupervised domain adaptation. The model excelled in processing complex datasets and outperformed current crop of classifiers.
A new model of the Earth's ionosphere has been developed using neural networks, which can reconstruct the topside ionosphere with high accuracy. This improvement is crucial for satellite navigation systems, such as global navigation satellite systems (GNSS), which require precise correction of radio signals to mitigate ionospheric delays.
Researchers reconstructing neurons from ctenophore nerve net discover a continuous neural network, challenging the neuron doctrine. The finding has potential to provide key information on the evolutionary origin of nervous systems.
Researchers found that exercise releases chemical signals that promote neuronal development in the hippocampus, a crucial area for learning and memory. Astrocytes play a critical role in mediating the effects of exercise on brain health, helping to regulate neuronal activity and prevent hyperexcitability.
A recent study found that humans process social situations similarly, with an extensive neural network processing various social information. The study identified main dimensions of social perception, including antisocial behavior, sexual or affiliative behavior, and communication.
Researchers at UTHealth Houston identified two brain networks involved in reading, working together to integrate word meanings. The study used electroencephalography recordings from patients with epilepsy to measure neural activity while reading complex sentences.
A team of UCF College of Medicine researchers created a digital topographical map of the cardiac sympathetic neural network, controlling heart rate and 'fight-or-flight' response. The map will guide treatments like neuromodulation therapy for hypertension, sleep apnea, and heart failure.
Scientists have developed a system called PhAST, which uses light-emitting enzymes and ion channels to transmit information between neurons. This method has shown promising results in restoring communication in defective circuits and modifying animal behavior.
Researchers Pawel Burkhardt and Fred Wolf will study the simple nervous system of marine organisms to understand brain evolution. The project aims to reveal the neural network of ctenophores, a predator with an alien-like brain.
A new method using a back propagation neural network improves the accuracy of orbit prediction and position error covariance prediction for space targets. The method reduces prediction errors by up to 170m compared to existing models.
A team of researchers has developed a new human-in-the-loop system to improve the accuracy and interpretability of deep neural networks. The system uses an interactive one-click method for annotating images, reducing the co-occurrence bias inherent in training datasets.
MIT researchers discovered unconventional activation functions that enable optimal neural network performance, leading to better classification on various datasets. The findings suggest that selecting the correct activation function can significantly improve data accuracy in machine learning applications.
Researchers developed a deep learning approach to recognize and predict motion using vector-based relative change in position. The method, VecNet+LSTM, scored higher than other frameworks in recognizing motion and predicting future movements. This study has implications for machine learning in video analysis and artificial intelligence.
Researchers at USC have developed a new type of chip with the best memory of any chip thus far for edge AI. The chip uses metal oxide memristors to store information in a compact and stable way, eliminating the von Neumann bottleneck in current computing systems.
Researchers developed a VR imaging system to measure neural activity in mouse brains during behavior, revealing abnormalities in cortical functional network dynamics associated with autism. The system successfully distinguished between autism model mice and wild-type mice based on their brain network patterns.
Researchers developed a novel design for the chip using a crossbar layout, outperforming state-of-the-art photonic counterparts in terms of scalability and technical versatility. The synergy of powerful photonics with the novel crossbar architecture enables next generation neuromorphic computing engines.
A new approach to describing network connections can help predict system strong and weak points, crucial for understanding disease spread and communication networks. Researchers found that mapping hierarchies and incoherence within a system enables prediction of strong and weak connections.
Researchers found that dogs with anxiety have altered brain connectivity, particularly between the amygdala and hippocampus. The study used fMRI to characterize abnormal neural networks in anxious dogs, providing insight into anxiety disorders in both animals and humans.
A recent study has revealed two distinct types of cells in the locus coeruleus, a small nucleus in the brainstem, connected via gap junctions. This discovery has far-reaching implications for understanding neuropsychiatric and neurodegenerative conditions.
Researchers developed a deep learning-based AI model to automate cirrhosis identification using large amounts of data from EHRs. The model successfully identified patients with cirrhosis with a precision of 97%, offering potential for early diagnosis and improved management of the disease.
A cross-disciplinary team developed a convolutional neural network to analyze microscopy images of chromosomes with cohesion defects. The algorithm achieved 73.1% accuracy in classifying new images, streamlining experiments with chromosome analysis.
Researchers at Columbia University are developing algorithms that enable robots to understand object permanence and learn from 3D information. This allows robots to track objects and humans as they move around, improving their perception capabilities in indoor environments.
Scientists at Max-Planck-Gesellschaft created an interactive atlas of gene expression in the zebrafish brain, revealing hundreds of genes with single-cell resolution. The new map integrates seamlessly with existing data, providing new insights into neural structure and function.
GPT-3 performs nearly on par with humans in decision-making but struggles with causal reasoning and information search. The language model's limitations may be due to its passive information-gathering approach, highlighting the need for active interaction with the world to achieve human-like intelligence.
A tiny worm called C. elegans is enabling scientists to explore the emerging theory that Parkinson’s disease starts in the gut, where a sticky protein causes neurons to die. The worms have a digestive tract and neurotransmitters similar to humans, making them a great model for studying cognitive problems associated with Parkinson's.
Thanhvu Nguyen will receive $510,509 from the National Science Foundation to develop NeuralSAT, a constraint-solving framework for verifying deep neural networks. The project aims to improve DNN verification and scalability, tackling challenges in precision and soundness.
The study reveals that engram neurons form new synapses during fear memory formation, while memory extinction leads to the disappearance of these connections. The researchers observed this process using a novel dual-eGRASP system, enabling longitudinal observations of identical synapses at multiple time points.
Researchers at University of Gothenburg developed AI method using graph theory and neural networks to analyze cell movement, enabling better understanding of biological processes and development of new medical technologies. The method can reconstruct cell paths and test medication effectiveness as potential cancer treatments.
Kyoto University researchers have created a map comparing circuit structure with neural activity in mammals, revealing a new mechanism behind visual cortex activities. This discovery sheds light on the hidden connections between neurons and could provide directions for constructing power-efficient deep neural networks.
Researchers at DZNE discovered that centrosome controls neuronal migration but not axon growth. The study used novel molecular tools to show that centrosomal activity influences radial migration of projection neurons.
Researchers have developed a novel experimental platform that combines optical and fMRI techniques to study large-scale brain networks. The platform identified the anterior insular cortex as a key player in controlling the Default Mode Brain Network, which is active during daydreaming, memory retrieval, and envisioning the future.
Researchers at Cornell University developed an AI tool to track online debates and detect when tensions are escalating. ConvoWizard provides real-time feedback to users on potentially incendiary language, encouraging constructive debate.
Researchers at MIT developed a technique to improve machine-learning models' reliability without requiring additional data or extensive computing resources. The method uses a simpler companion model to estimate uncertainty, enabling more effective uncertainty quantification.