Articles tagged with Neural Modeling
A team from Nagoya University created an artificial neural network model that performed the delayed matching-to-sample task and analyzed its behavior. The model was able to evolve to exhibit human-like metamemory, adapting to its environment by learning and evolving. This breakthrough aims to create machines with memories like humans.
Researchers at Sainsbury Wellcome Centre found that mice can choose the best escape route after only 10 minutes of exploration, without needing to experience threat. The study suggests that mice use innate heuristics and natural exploration to learn this information.
Researchers have successfully processed sequences with a large neural network while consuming significantly less energy on neuromorphic hardware. This breakthrough showcases the potential of neuromorphic technology to improve the energy efficiency of AI workloads.
Researchers at EPFL's School of Life Sciences create a digital twin of Drosophila called NeuroMechFly, which uses biomechanical modeling and machine learning to simulate the fly's movements. The model is validated through experiments that demonstrate its accuracy in replicating real animal behaviors.
Researchers developed a deep learning-based model to predict drug-drug interactions using gene expression data. The DeSIDE-DDI model can identify potentially dangerous pairs and act as a drug safety monitoring system, helping establish the correct usage of drugs in the development phase.
Researchers developed a machine-learning method that allows robots to pick up and place never-before-seen objects in random poses, requiring only 10 human demonstrations. The system uses a neural network specifically designed to reconstruct 3D shapes, enabling the robot to generalize to new object orientations.
Researchers at Salk Institute discover that brain parses information through interactions of waves of neural activity, changing how data is processed and affecting attention and focus.
The VIGICOVID system uses deep neural architectures to answer natural language questions about COVID-19 and SARS-CoV-2. Researchers have developed a prototype that provides good results and can be easily scaled up for marketability.
Researchers at Kaunas University of Technology improved an algorithm to detect Alzheimer's disease from MRI images, achieving over 98% accuracy. The new model uses a modified neural network and adapts to variations in data, such as differences in hospital equipment and patient positions.
A new mathematical framework has been created to study fitness landscapes of regulatory DNA, enabling the prediction of gene expression changes. The framework uses a neural network model trained on millions of experimental measurements to decipher the evolutionary past and future of non-coding sequences.
Researchers have created a groundbreaking dataset of ultrasonography scans of three major arteries supplying blood to the brain in children. The dataset consists of 821 participants, allowing for the development of machine learning models that can accurately predict a child's age and cognitive abilities based on their ultrasounds.
Researchers at the Medical College of Wisconsin uncovered that conceptual knowledge is tied to perceptual and experiential information. They used fMRI to measure neural activity while participants read hundreds of words, finding that experiential information was key to understanding word meaning.
Adversarially robust models capture aspects of human peripheral processing, with results showing similarity in image transformations and perception alignment. The study's findings shed light on the goals of peripheral processing in humans and could help improve machine learning models.
GIST researchers propose a new strategy for crime prevention using artificial intelligence, trained on a large-scale dataset of deviant incident reports and corresponding images. The model, called DevianceNet, can accurately classify and detect deviant places, making it a useful tool in urban safety development.
Researchers studied how diverse neural network training datasets impact generalization. They found that data diversity is key to overcoming bias, but also degrade performance when neural networks are trained for multiple tasks simultaneously. The study highlights the importance of designing diverse and controlled datasets in machine le...
Researchers at Tokyo Institute of Technology have developed a new AI processor called Hiddenite, which achieves state-of-the-art accuracy in sparse neural networks with lower computational burdens. The chip drastically reduces external memory access for enhanced computational efficiency.
The MIT team developed a computer model that can perform sound localization tasks as well as humans, and adapts to real-world environments. The model uses convolutional neural networks and was trained on over 400 sounds, including human voices and animal sounds.
A team of scientists developed an AI-based model to predict personal thermal comfort based on spatial parameters, achieving exceptional accuracy. The study highlights the importance of incorporating architectural features in models to reduce energy consumption.
MIT researchers develop a method to test feature-attribution methods for machine-learning models. They find that even the most popular methods often miss important features in an image and some perform as poorly as a random baseline. This has major implications for high-stakes situations like medical diagnoses.
Researchers developed a neural network model called STANN that provides new insights into the brain's cellular architecture and functionality. The model predicts precise locations of different cell types and their communication patterns within morphological layers.
Researchers found that re-identifying individuals from genomic data using public face images is harder than previously thought, with success rates well below idealized settings. They developed a method to alter social media photos and reduce the risk of privacy breaches.
A multidisciplinary team from UCI has published a new guide for statistical analysis in neuroscience research, providing an overview of mixed-effects models and clear instructions on their application. The guide aims to improve the validity and reproducibility of experimental findings in the field.
Researchers trained an artificial intelligence algorithm to predict the next designer drugs before they are even on the market, allowing law enforcement agencies to identify and regulate new versions of dangerous psychoactive drugs. The model was tested against 196 new designer drugs and found nearly all were present in its generated set.
A new recursively embedded atom neural network (REANN) model improves material simulation accuracy by incorporating local completeness and nonlocality. The model outperforms current machine learning models in describing the local environment, enabling more accurate predictions.
Researchers developed AfriBERTa, a neural network model that achieves state-of-the-art results for low-resource African languages. The model works with 11 languages spoken by over 400 million people and requires only one gigabyte of data, compared to thousands for existing models.
Researchers at Skoltech propose a method for interpreting brain activity data that is up to five times more accurate than the conventional technique. This method can help treat drug-resistant epilepsy and understand cognitive processes in the healthy brain.
A new study reveals that high-performing AI next-word prediction models resemble the function of language-processing centers in the human brain. The models' activity patterns closely match those seen in the brain during language tasks, suggesting a potential connection between AI and human language processing.
A team of scientists at NAIST successfully used automatic differentiation to accelerate calculations of model parameter extraction, reducing computation time by 3.5 times compared to conventional methods. This breakthrough enables the design of more efficient power converters with increased performance and reduced energy consumption.
A novel machine learning approach has been developed to understand symmetry and trends in materials, enabling researchers to group similar classes of material together. The technique uses a large, unstructured dataset gleaned from 25,000 images to identify structural similarities and trends.
A study published in Communications Biology reveals that chronic hyperglycemia in diabetes impairs working memory performance by altering the connection between key brain regions. Researchers found that areas critical for forming and retrieving memories were over-connected, leading to errors in remembering correct information.
A neural model suggests that conspiracy theories are a result of the brain's tendency to simplify complex information and associate it with emotional experiences. This theory challenges traditional views on the origins of conspiracy theories, proposing a link between genetic determinism and neuronal activity.
Researchers developed a robust, deep neural network model to analyze automobile traffic impacts of construction zones. The model estimates hourly traffic volumes without adjustment factors, helping transportation agencies plan for efficient work zone operations.
A deep learning-based method developed by Kaunas University of Technology researchers can predict the possible onset of Alzheimer's disease from brain images with an accuracy of over 99%. The algorithm was trained on functional MRI images from 138 subjects and performed better than previously developed methods.
Researchers at Tokyo Institute of Technology developed a tunable neural network framework that achieves high accuracy and efficiency for sparse CNNs. The new architecture employs a Cartesian-product MAC array and pipelined activation aligners to enable dense computing of sparse convolution, resulting in better resource utilization.
Neuroscientists used artificial intelligence to disentangle the relationship between perception and memory in the human brain. A novel computational framework predicts neural responses in the primate visual system, resolving decades-long debates over the role of the medial temporal lobe (MTL) in perception.
A new neural model can detect 'inappropriateness' in chatbot messages, which are defined as content that may harm the reputation of the speaker's company without being toxic. The study provides a large collection of labeled datasets for further research and has been made publicly available.
The brain's globally sparse yet locally compact modular topological characteristics reduce resource consumption for establishing connections. The research model shows that rewiring the network to a more biologically realistic modular structure significantly reduces running consumption and building cost.
A triple-system neural model of maladaptive consumption presents a new understanding of compulsive seeking and consumption behaviors. The model identifies three brain regions that may drive maladaptive behaviors: the impulsive system, reflective system, and interoceptive-awareness system.
Researchers have created a theoretical model that accurately predicts the neural activity of a mouse brain without requiring fine-tuning. The model uses critical phenomena to explain phase transitions in physical systems and may have applications for studying complex dynamical systems.
Researchers developed a new analysis system to study brain function and diseases using 3D artificial brain models. The system allows for precise non-destructive stimuli and real-time measurement of neural signals, providing insights into functional connections between brain cells.
Researchers found that native Russian speakers can precisely predict specific words and grammatical properties of words, with neural network models showing comparable precision. The study also discovered that the neural network predicts low-probability words better than humans and predicts high-probability words worse than humans.
A new study from the University of Bern has developed a simplified computational model to accelerate brain research. The approach uses a mathematical relation to reduce complexity while retaining accuracy, allowing for easier characterization and simulation of large neural networks.
Researchers developed an artificial neural network model that can simulate brain processes involved in grasping movements. The model was trained with data from rhesus monkeys and accurately reproduced their grasping movements, providing insights into neuronal dynamics.
Researchers discovered that the anterior cingulate cortex is involved in using mental models to simulate the future and make decisions. The brain structure encodes multiple aspects of decision-making, including the likelihood of a specific outcome.
New research from Cold Spring Harbor Laboratory highlights the importance of model evaluation in neuroscience. By building and comparing several models of neural signaling, researchers found that good predictive power does not necessarily indicate a model's representation of real neural networks.
A study from MIT neuroscientists suggests that parts of the brain originally evolved for object recognition have been repurposed for reading. In nonhuman primates, the inferotemporal cortex is capable of distinguishing words from nonsense words and picking out specific letters.
Researchers propose a new 'Reciprocal Inhibition Model' of PTSD, explaining inhomogeneous symptoms by addressing variabilities at neural, attentional, and symptom levels. This model may aid tailored clinical treatment for individual patients.
A team of researchers from NYU Tandon and Canada have developed a machine learning model called PHTNet, which enables robots to accurately predict and compensate for hand tremors in patients with Parkinson's disease. The model has been tested on a dataset of 81 patients and reported a 95% confidence rate over 24,300 samples.
Researchers at MIT have developed a machine learning method to fill in the missing low-frequency seismic waves in human-generated seismic data, allowing for more accurate mapping of underground structures. The technique was trained on simulated earthquakes and used to infer missing frequencies from new input data.
A deep neural network model developed by Finnish researchers can predict healthcare visits among elderly people with high accuracy. The model has the potential to save millions of dollars in unnecessary funding for healthcare services.
Researchers develop software that mimics human visual processes to improve color accuracy, outperforming existing methods in psychophysical tests. The new framework reduces and extends the color gamut, producing results free from artefacts and meeting the film industry's demands.
MIT researchers create a visual deprojection model that recovers valuable data lost from images and video by learning patterns in low-dimensional projections. The model has successfully recreated video frames showing people walking and recovered motion-blurred images, with potential applications in medical imaging.
Researchers developed a neural network model using machine learning to predict Universe structure formation. The new model is more accurate than existing analytic methods and efficient enough for large-scale simulations.
Researchers at Columbia University have developed a new technique using reinforcement learning to train a neural network model to plan synthetic routes to any target molecule. This approach is more successful than existing strategies and can optimize user-specified objectives such as cost, time, and sustainability.
Paul Sajda, a professor at Columbia University, has been awarded the Vannevar Bush Faculty Fellowship for his research on cognitive neuroscience. The fellowship will support his project to develop a testable model of human brain dynamics governing rapid decision-making in natural environments.
Researchers at MIT have demonstrated that artificial neural networks can be used to drive specific brain neurons, showing a strong activation pattern. The study suggests that these models could be used to control brain states in animals and establish their usefulness, paving the way for further research.
Researchers developed a computational model that explores how the auditory system achieves accurate speech recognition by identifying distinct categories of sounds. The model found that the brain looks for informative features, such as those characteristic of a face, to distinguish between different vocalizations.
Researchers created a new technique called neural source-filter (NSF) to synthesize high-quality speech waveforms resembling the human voice. NSF requires less data and parameter tuning compared to existing methods, resulting in comparable quality to WaveNet.
The new algorithm simplifies neural models through synaptic pruning and dendritic pruning procedures, resulting in simplified structures that can be implemented as logic circuits. These circuits achieved satisfactory classification accuracy on benchmark problems, suggesting potential for solving complex real-world problems with high ha...
A user-friendly software tool models neural circuits in outer brain layers, enabling EEG-guided treatment for patients based on new knowledge of underlying neural circuits. The Human Neocortical Neurosolver is a free, open-source tool that can help bridge the gap between genetic and molecular changes to neural circuit level signals.