Articles tagged with Neural Networks
Scientists developed a generative neural network to create new pharmaceutical medicines with specific properties. The network, trained on millions of molecular structures, identified 69 potential anticancer compounds and hundreds more using a powerful extension of the method.
Researchers discovered that an algorithm called additive increase, multiplicative decrease (AIMD) is used both in engineered systems like the Internet and biological networks like the human brain. This finding sheds light on how the brain manages information and potentially helps understand learning disabilities.
Neuroscientist David Heeger proposes a new theory that explains how the brain uses prediction and inference to make decisions. The brain's neural network can process sensory input in a feedforward manner, but also runs in a feedback mode to generate predictions, and combines both modes for optimal performance.
Researchers discovered two distinct neural networks controlling the balance between speed and accuracy when making decisions. The subthalamic nucleus region plays a key role in this process, with one network requiring more information for accurate decisions and another lowering the threshold for quick choices.
Researchers used rabies viruses to visualize neural transplant integration in mouse brains, revealing region-specific connections between transplanted cells and host neurons. The approach opens new prospects for predicting and optimizing the ability of neural transplants to functionally integrate into a host nervous system.
A new computational model reveals a winner-take-all operation is enabled by a configuration of inhibitory neurons. The model, developed by MIT researchers, uses theoretical computer science to prove that a specific arrangement of inhibitory neurons is necessary for the operation.
Researchers found that inhibitory brain cells form maps that broaden with maturation, unlike excitatory neurons which refine and define areas. This discovery sheds light on how the brain organizes and processes information.
Researchers at MIT's CSAIL have developed a new system that analyzes correspondences between images and spoken descriptions to train speech-recognition systems. The system can potentially provide automatic speech recognition for less-resourced languages, leading to fully automated translation capabilities.
Research reveals long projecting neurons coordinate fore- and hindlimb movements, maintaining stability and rhythm. Elimination of these neurons impairs running speed and coordination.
Researchers at U of T Engineering developed an AI algorithm that learns directly from human instructions, exceeding conventional training methods by 160% and outperforming its own training by 9%. The algorithm's potential lies in applying heuristic training to fields like medicine and transportation.
Researchers have developed a new method to record brain activity in living mice, capturing the dynamic activity of thousands of neurons in three dimensions. The technique, known as 'light sculpting,' uses laser pulses to illuminate and analyze the activity of neurons within specific layers of the brain.
Researchers at MIT's CSAIL have developed a new way to train neural networks that provide not only predictions and classifications but also rationales for their decisions. The system consists of two modules: one extracts segments of text from training data and scores them, while the other performs prediction or classification tasks.
Researchers successfully integrated transplanted embryonic nerve cells into the visual cortex of adult mice, demonstrating functional connectivity and restoration of network function. This breakthrough holds promise for treating acquired brain diseases, including neurodegenerative illnesses and stroke-induced damage.
A new technology, developed by University of Calgary researchers, enables recording brain cell activity for weeks with higher resolution than conventional methods. This allows researchers to investigate neurological diseases and cognitive functions like learning and memory in animal models.
Researchers discovered that bilinguals use distinct neural networks to read languages with phonetic and orthographic correspondence, such as Basque, from those without it, like English. The findings have significant implications for teaching reading to adults and children.
A study using MRI found significant structural differences in the brains of children with PTSD compared to those without. The findings suggest a shift towards a more localized network structure, which may be a target for future treatments. This research could help develop new interventions for pediatric PTSD.
A new tool developed at Lawrence Berkeley National Laboratory enables researchers to interactively explore brain hierarchical processes and shed light on neurological diseases like Alzheimer's. Brain Modulyzer combines multiple views of functional magnetic resonance imaging (fMRI) data to provide context for brain connectivity data.
Researchers at the Universities of Dundee and Strathclyde have identified a mechanism that allows neurons to protect against spreading brain damage. The discovery, published in Scientific Reports, suggests that stimulating this network activity could limit major brain damage and shorten recovery periods.
A recent study published in Scientific Reports found that language learning boosts brain activity and increases neuroplasticity, leading to improved information processing. The more languages a person knows, the faster their brain adapts to new information.
A new app developed by Cornell researchers uses artificial intelligence to identify furniture brands and retailers based on photos of products. The system was trained using crowdsourced images and can search a vast database of iconic images from manufacturers' catalogs or specialized websites.
A novel imaging technique allows visualization and monitoring of structural alterations in neuronal networks after traumatic brain damage, stroke, or aging processes. This enables detection and characterization of diseases like dementia, epilepsy, and metabolic disorders.
Researchers demonstrate that groups of activated neurons can form the basic building blocks of learning and memory. They used optogenetic tools to control and observe brain activity in living mice, finding that neural ensembles can be artificially implanted and replayed.
Researchers at Salk Institute discover that the timing of brain activity, not just the number of spikes, is crucial for recognizing shapes and perceiving the world. The study's findings have potential applications in developing more accurate visual prosthetics for people with blindness.
A new study reveals an evolutionary universal brain structure that enables comparisons of cortical networks between species. This common architecture provides insights into brain disorders such as Alzheimer's and schizophrenia.
Researchers created a mini-brain model to study idiopathic autism, revealing early neuronal overgrowth and dysfunctional cortical networks. The model shows a defective Wnt pathway and misregulated neurotransmitters, leading to reduced excitatory synapses and functional defects.
Recent breakthroughs in creating artificial systems that outplay humans in games are rooted in neural networks inspired by information processing in the brain. The complementary learning systems theory explains how humans and animals learn, highlighting its potential importance as a framework for AI development.
Researchers identified 'stop cells' in the brainstem of mice and lampreys that quickly end movement by activating neural networks. The study provides new insights into the neuronal control of movement termination in vertebrates.
Scientists have discovered a special neural network in the primate brain that anticipates all possible situations, allowing for novel behavior adaptation. This 'reservoir computing' property enables the brain to create a universal representation of combinations, preparing primates for unlimited situations.
New research reveals that hierarchy in biological networks arises due to cost constraints on connections, leading to more efficient networks. This finding may accelerate the development of complex computational brains in AI and robotics.
Researchers from KU Leuven have developed machines that can learn to tell the difference between familiar and unfamiliar objects, such as a blurred shape on the road. This technology has the potential to improve self-driving car safety by allowing machines to make more reasonable decisions in new situations.
Researchers at MIPT create electronic synapses based on HfO2 memristors, exhibiting properties similar to biological synapses. The devices can model complex learning mechanisms, including LTP and LTD, and demonstrate spike-timing-dependent plasticity.
Researchers at Numenta Inc. have published a new theory on how networks of neurons in the neocortex learn sequences, providing a breakthrough in understanding neural circuits.
Researchers at UCSB have mapped the network of circadian neurons that communicate to re-establish synchronization, finding a 'small-world structure' with hubs and short paths for communication. This discovery sheds light on how the suprachiasmatic nucleus (SCN) regulates essential functions like sleep and hormone release.
Researchers have mapped the largest network of cortical neurons to date, shedding light on how brain circuits are organized. The study reveals modular architecture and functional specific connectivity between neurons, enabling reverse engineering of the brain's structure and function.
Researchers developed a 3D micro-scaffold technology that promotes reprogramming of stem cells into neurons and supports growth of neuronal connections. The system improved cell-survival rates by nearly 40-fold compared to individual cell injections, enabling the potential treatment for human neurodegenerative disorders.
The Allen Institute will reconstruct over 1,000 brain connections using electron microscopes and machine learning algorithms. The project aims to advance artificial intelligence by reverse-engineering the brain's algorithms.
A new learning procedure developed by Robert Gütig enables neural networks to filter out irrelevant sensory impressions and link them to events occurring after a delay. This breakthrough has significant implications for technological applications like automatic speech recognition.
Researchers at Max Planck Florida Institute for Neuroscience used electrophysiological and optical approaches to visualize and manipulate neuronal activity in individual neurons of the somatosensory cortex. They found that the formation of functional microcircuits was determined by specific settings and the number of neurons stimulated...
A new study suggests that specific brain network alignment influences whether altruism stems from empathy or reciprocity. Researchers found distinct patterns in brain connectivity between individuals exhibiting different levels of altruism.
A UCLA-led collaboration has identified a specific gene network and pattern of expression that promotes repair in the peripheral nervous system. This network does not exist in the central nervous system, but researchers have found a drug that can enhance nerve regeneration there.
Researchers at Virginia Tech Carilion Research Institute found that brain cells change through experiences like learning and emotions, forming functional networks.
Fruit flies develop complex nervous systems through a set of genetic control switches that interact early in development to generate dozens of types of olfactory neurons. The same gene network also plays a role in programming taste neurons, suggesting the same basic mechanism could be at work in other animals.
Researchers found that early nerve cell development creates brain circuit templates, which are then disrupted by trauma, leading to delayed effects on thoughts and memories.
Researchers at MIT have developed a new chip called Eyeriss that can enable mobile devices to run powerful artificial-intelligence algorithms locally. This could improve performance, reduce latency and allow for more efficient processing of data without relying on Wi-Fi connections.
Researchers found that HIV protein Tat alters networked neuron activity and leads to adaptations that improve survival but impair function. The study suggests that targeting these adaptations could facilitate therapeutic intervention for seizure disorders and other neurological symptoms in HIV-infected patients.
Researchers from Russia and Italy develop neural network based on polymeric memristors, enabling machine vision and intelligent control systems. The networks can learn and perform logical operations, offering a promising alternative to traditional computing methods.
Lab-grown neural networks can replace damaged axon tracks in patients with severe head injuries, strokes, or neurodegenerative diseases. The micro-TENNS, made of mature cerebral cortical neurons and long axonal projections, integrate into existing brain structures and reconstitute missing pathways.
Researchers discovered that 70% of brain information passes through only 20% of neurons in cortical regions, suggesting a vital role in communication and learning. This high-traffic network is also more vulnerable to disruption, which could impact brain health.
Researchers have developed a human in vitro cellular model of Cockayne syndrome, a devastating neurodegenerative condition. The model reveals key aspects of the disease, including altered electrophysiological activity and impaired neuronal function.
Researchers at Case Western Reserve University created a complete model to describe serotonin's role in brain development and structure. The study supports the importance of serotonin for normal functioning of neurons, synapses, and networks in the cortex.
Darwin is a neuromorphic hardware co-processor developed by Chinese researchers that enables efficient execution of Spiking Neural Networks in resource-constrained embedded systems. It supports flexible configuration and has potential applications in intelligent hardware systems, robotics, brain-computer interfaces, and more.
Researchers found that humans can recognize objects even when only a small portion is visible, and validated an algorithm to explain human learning. The algorithm can be used for machine learning, data analysis, and computer vision to improve performance.
Research reveals LSD reduces connectivity within brain networks, disrupting synchronization of neural firing patterns. This change affects the underlying thought and behavior processes in humans.
Researchers found joiner neurons that can be pre-positioned for rapid recruitment into memories, changing how brain networks adjust to build memories. This discovery has implications for understanding memory formation and developing new strategies for brain injury recovery.
Leading US neuroscientists advocate for a coordinated national network of neurotechnology centers to accelerate the BRAIN Initiative. The proposed brain observatories would unite and synergize hundreds of individual laboratories, enabling rapid progress in key areas like connectomics and neural nanoprobe systems.
A national network of neurotechnology centers is proposed to accelerate the BRAIN Initiative, a decade-long scientific project with over $300 Million budget. The centers would tackle critical areas like Connectomics, Nanotechnology, and Optical Imaging, providing an alternative STEM career path for postdocs and graduate students.
Researchers discovered two major types of cortical neurons providing separate mechanisms for stimulus-specific adaptation in the brain. This phenomenon helps the brain distinguish between expected and unexpected sounds, enhancing sensitivity to rare sounds. The findings provide new insights into the neural basis of key hearing phenomena.
Rice University researchers are combining experiments and computational analysis to learn how the brain organizes itself after injury. They aim to direct the growth of new neurons to treat stroke and neurodegenerative diseases.
Researchers have identified kisspeptin neurons as the key players in generating episodic hormone pulses crucial to normal reproductive functioning. These findings hold promise for developing new fertility treatments targeting brain circuitry disorders, which contribute to up to one-third of all infertility cases.
University of Oregon scientists have identified a crucial network of neurons in fruit fly larvae that mirrors the human nervous system. This discovery may lead to innovative treatments for motor disorders and more efficient robotic systems.