Articles tagged with Neural Networks
Researchers identified new neurons that respond to different spatial frequencies, allowing for more precise object recognition, and used digital twins to confirm the findings in mouse brains.
Researchers developed photonic computing chips that enable fast, all-optical learning and decision making, overcoming key limitations for photonic spiking neural systems. The new chips could improve autonomous driving technologies and enable robotic systems that learn through real-world interactions.
Researchers developed a new method called Learn-to-Steer, which analyzes internal attention patterns of image-generation models to guide their placement according to user instructions. The approach improved accuracy in understanding spatial relationships by up to 61% in existing trained models.
Researchers found that dosed nonlinearity improves model performance in various tasks, especially with limited data. Nonlinear units function like flexible switches, adapting linear processing modes based on context.
Researchers created a new frequency-aware approach to crafting adversarial images that better match human visual perception. The method, called Input-Frequency Adaptive Adversarial Perturbation (IFAP), significantly outperformed existing techniques in structural and textural similarity.
Engineers at the University of Pennsylvania have discovered that foams exhibit internal motion resembling deep learning in AI systems. The study suggests a common mathematical principle underlying both foams and AI training, with implications for designing adaptive materials and understanding biological structures.
Researchers at Rockefeller University have made a breakthrough in understanding how the brain controls facial expressions, discovering a complex network of neural circuits involved. Contrary to long-held assumptions, both lower-level and higher-level brain regions are involved in encoding different types of facial gestures.
Researchers reveal that harm causes stronger guilt, while sense of responsibility triggers shame, influencing compensatory behaviors. The study also identifies distinct neural activity for guilt and shame-driven decisions, shedding light on the cognitive processes guiding these emotions.
This study applies Physics-Informed Neural Networks (PINNs) and Extreme Learning Machines to solve complex option pricing problems under stochastic volatility. The research enables accurate pricing of American-style options for both equity and real estate index derivatives, addressing a significant challenge in quantitative finance.
A research team developed an AI-guided framework to discover new metallic glasses by combining element embeddings learned from Wikipedia with graph neural networks. This approach overcomes challenges in predicting glass-forming systems, enabling the discovery of promising compositions with high glass-forming ability.
Researchers at Institute of Science Tokyo developed a neural-network-based 3D imaging technique that can precisely measure moving objects. The new method reconstructs high-resolution 3D shapes using only three projection patterns, enabling dynamic 3D measurement across various applications.
KVzip reduces chatbot response time and memory cost while maintaining accuracy, achieving 3–4× memory reduction and approximately 2× faster response times. The technology also demonstrates scalability to extremely long contexts and has been integrated into NVIDIA's open-source library.
Autograph, a new framework, uses graph neural networks and deep reinforcement learning to achieve higher accuracy and faster execution of compute-intensive programs. It outperformed other approaches across various datasets, with notable improvements on Polybench, NPB, and SPEC 2006 benchmarks.
Recent advances in deep learning techniques have overcome limitations in spatial and temporal resolution of BOLD-fMRI. DL models improve image quality through super-resolution reconstruction, automate segmentation, and enhance registration, enabling finer localization of neural activity and more precise brain activity quantification.
Researchers developed a machine learning-based workflow, SPaDe-CSP, to predict crystal structures of organic molecules. The workflow narrows the search space by predicting probable space groups and crystal densities before computationally intensive relaxation steps.
A team of researchers at the University of Waterloo developed a framework that uses mathematical tools and machine learning to rigorously check and verify the safety of AI-driven systems. The framework has been tested on challenging control problems and matched or exceeded traditional approaches.
A team of researchers from Yokohama National University has developed a novel compact superconductive neuron device that operates at high speeds with ultra-low power consumption. The device eliminates variation in elemental circuit characteristics, achieving ideal input-output characteristics and resolving the vanishing gradient problem.
VFF-Net applies label-wise noise labelling, cosine similarity-based contrastive loss, and layer grouping to improve image classification performance compared to conventional forward-forward networks. The algorithm reduces test errors on various datasets, enabling lighter and more brain-like training methods that make AI more sustainable.
Salk scientists pinpoint gracile nucleus as brain area responsible for differentiating between painful and non-painful touch, with dysfunction leading to chronic pain. Altered neuronal activity in the dorsal column nuclei drives mechanical allodynia, causing the brain to misinterpret innocuous light touch as painful.
Researchers have identified a new population of hypothalamic neurons, Crabp1 neurons, that play a critical role in regulating energy expenditure. Silencing these neurons leads to reduced energy expenditure and obesity, while activating them enhances locomotor activity and protects against high-fat diet-induced weight gain.
Researchers have developed an emulator called Effort.jl that mimics the behavior of large-scale structure models, allowing for fast analysis on standard laptops. The new model delivers similar accuracy as the original, enabling scientists to analyze upcoming data releases from experiments like DESI and Euclid.
The team will study neurons within a brain organoid, a millimeter-sized, three-dimensional structure grown in the lab from adult stem cells, to design smarter and more sustainable artificial intelligence. They aim to replicate complex computations that occur in the human brain to improve AI efficiency.
Researchers at Tohoku University have developed an AI-built materials map that combines experimental data with computational predictions to identify promising materials for thermoelectric waste-heat recovery. The map enables faster development timelines and reduces trial-and-error, accelerating innovation in energy-related technologies.
Researchers developed a simple model that reproduces deep neural network features, allowing for optimized parameter tuning. The 'folding ruler' model demonstrates how nonlinearity and noise improve network performance, enabling more efficient training without trial-and-error.
A new photonic neural network developed in China achieves higher classification accuracy than digital models by using physical light transformations and multisynaptic optical paths. The system's design avoids errors introduced by translating software to hardware, marking a major step forward in optical AI hardware.
Researchers used Bayesian neural network to identify relationships between gut bacteria and metabolites, providing clues about health. The approach outperformed existing methods in analyzing sleep disorder, obesity, and cancer studies.
Researchers have successfully controlled a dexterous robotic hand using noninvasive EEG-based Brain-Computer Interfaces (BCIs) for individual finger movements. The study demonstrates real-time brain decoding and motor imagery control, paving the way for potential applications beyond basic communication to intricate motor control.
A new deep learning model, CNN-SENet, leverages GNSS-R data to improve wind speed retrieval. The model outperforms conventional models in both speed and precision, offering promising tool for global ocean wind monitoring.
Researchers identified a key brain region, the ventrolateral periaqueductal gray matter (vIPAG), that modulates pain and emotional responses in threatening situations. By inhibiting specific cells within this region, scientists discovered a potential pathway for better pain relief.
The study proposes an event-triggered asymptotic composite neural tracking control scheme for intelligent vehicles, addressing system nonlinearities and uncertainties. It enhances tracking precision and reduces communication traffic through variable threshold-based triggering conditions.
The Rice University team created a soft robotic arm capable of performing complex tasks using smart materials, machine learning, and an optical control system. The arm is guided and powered remotely by laser beams without any onboard electronics or wiring.
A revolutionary AI model has been developed to diagnose lung cancer without relying on costly GPU servers or massive datasets. The ultra-lightweight model achieved a discrimination performance corresponding to an AUC value of 0.92, outperforming state-of-the-art large-scale AI systems.
Researchers found that AI chatbots have difficulty detecting adverse drug reactions and providing personalized advice. The study suggests that improving AI for mental health concerns could be life-changing for communities with limited access to healthcare.
Researchers employed Bayesian neural networks to fit photonuclear cross-sections with remarkable reliability, outperforming traditional methods like TENDL-2021. The approach demonstrated superior accuracy in describing low-energy thresholds and high-energy tails, particularly for sparse or biased data.
Researchers at Johns Hopkins University found that AI systems struggle to understand social dynamics and context necessary for human interaction. Human participants were able to accurately rate features important for understanding social interactions, while AI models failed to match human brain and behavior responses across the board.
The AChemS 47th Annual Meeting features cutting-edge research on chemosensory perception, including taste and smell dysfunction in cancer patients and potential associations with learning and memory decline. The conference also highlights the impact of GLP-1 Receptor Agonists on human taste ability.
A team of researchers developed Lp-Convolution, a novel method that uses multivariate p-generalized normal distribution to reshape CNN filters dynamically. This breakthrough improves the accuracy and efficiency of image recognition systems while reducing computational burden.
A Lehigh University team developed a novel machine learning method to predict abnormal grain growth in materials, enabling the creation of stronger, more reliable materials. The model successfully predicted abnormal grain growth in 86% of cases, with predictions made up to 20% of the material's lifetime.
Researchers have created a breakthrough photonic chip that can train nonlinear neural networks using light, accelerating AI training while reducing energy use. The chip uses a special semiconductor material to reshape how light behaves, enabling reconfigurable systems with wide mathematical function expression.
Scientists have developed an all-optical activation function based on sound waves for photonic computing, enabling the creation of energy-efficient artificial intelligence systems. This breakthrough could potentially facilitate the scaling up of physical computing systems and pave the way for more efficient optical neural networks.
A new hardware platform for AI accelerators capable of handling significant workloads with reduced energy requirement has been developed. The platform leverages III-V compound semiconductors to create photonic integrated circuits, which operate at the speed of light with minimal energy loss.
A new USC-led study using fMRI reveals the neural mechanisms that contribute to urinary incontinence in stroke survivors. The research found significant differences in brain activity during voluntary versus involuntary bladder contractions, presenting potential pathways for targeted therapies.
Researchers have identified specific brain structures that regulate political passion, finding damage to the prefrontal cortex increases intensity and amygdala decrease it. The study suggests emotion plays a role in shaping expressed political beliefs rather than determining ideology.
A new study from USC Dornsife's Brain and Creativity Institute found that nostalgic music engages the brain's default mode network linked to memory and self-reflection, as well as its reward circuitry. This discovery could support emotional well-being and cognitive function in individuals with memory impairments.
Distributed acoustic sensing systems face data processing speed limitations; researchers leverage photonic neural networks to overcome these challenges. The TWM-PNNA system achieves high recognition accuracy above 90% with low power consumption, outperforming electrical GPUs by orders of magnitude.
Researchers at Saarland University are developing leaner, customized AI models and techniques like knowledge distillation to reduce energy consumption. These smaller models enable small and medium-sized businesses to access powerful AI technology without a large technical infrastructure.
A new neural network can identify fish activity on coral reefs by sound, faster than human experts, enabling real-time monitoring of fish populations, species identification, and disaster response. This technology has the potential to revolutionize ocean monitoring and research.
Researchers at Technical University of Munich developed a new AI training method that significantly reduces energy consumption. The approach uses probabilities to determine parameters, making the training process 100 times faster while maintaining accuracy comparable to existing procedures.
Researchers have identified three cell types in the median raphe nucleus that control decisions on perseverance, exploration, and disengagement. These findings may help understand neuropsychiatric conditions such as OCD, autism, and major depressive disorder.
Researchers found differences in genes and brain wiring between forest and desert flies, explaining how climate change impacts insects. Forest flies show increased avoidance of heat, while desert flies are attracted to warmer temperatures.
Researchers discovered a horizontally distributed and modular organization of cortical movement units, with different types of neurons forming functional clusters in distinct regions. The study also found that the brain re-networks and adapts to learn new motor skills.
Researchers develop a novel adaptive nonlinear PID controller integrated with radial basis function neural network for enhanced ballbot functionality. The proposed NPID-RBFNN controller demonstrates superior stability and robustness, outperforming traditional PID and NPID controllers.
Researchers created a computational method to track brain cell development over time, capturing unlabeled cells and fine structures in live cultures. The algorithm achieved high precision rates for detecting individual neurons, paving the way for studying neurological diseases and developing therapies.
A study by Nagoya University researchers found that excessive neuronal activation over time leads to brain function decline, contradicting previous theories. Interventions targeting reduced neuronal hyperactivation, such as dietary changes, may mitigate age-related cognitive decline in humans.
Harvard researchers have developed a silicon chip capable of recording small yet telltale synaptic signals from a large number of neurons. The chip has successfully mapped over 70,000 synaptic connections from approximately 2,000 rat neurons.
A new study led by researchers at Mass General Brigham suggests that different brain regions activated by creative tasks are part of one common brain circuit. People with brain injuries or neurodegenerative diseases may have increased creativity due to changes in this circuit.
The study compared the performance of seven AI models with that of 400 humans in comprehension tasks, revealing a significant difference in accuracy. Human accuracy reached 89%, while AI models struggled to achieve more than 70% correct answers.
A new optical encryption system uses holograms and neural networks to encode information, making it virtually unbreakable. The system achieves an exceptional level of encryption by utilizing a neural network to generate the decryption key.
A recent study published in JAMA Network Open found that heavy cannabis users exhibited reduced brain activity during working memory tasks, associated with worse performance. Abstaining from cannabis before cognitive tasks may help improve performance.
The new model, based on a PV-RNN framework, achieves compositionality by combining language with vision, proprioception, working memory, and attention. It requires less computing power than large language models (LLMs) and makes mistakes similar to humans.