All articles tagged with #deep learning
Originally Published 5 days ago — by Nature
The article introduces SleepFM, a large-scale foundation model trained on over 585,000 hours of sleep data from 65,000+ participants, which captures complex sleep physiology across multiple modalities and demonstrates strong predictive power for a wide range of diseases, outperforming traditional models and showing robust generalization across datasets and time.
Originally Published 27 days ago — by MIT News
MIT engineers developed a deep-learning model that predicts, with 90% accuracy, how individual cells in a fruit fly embryo fold, divide, and rearrange during early development, potentially aiding in understanding tissue formation and early disease detection.
Originally Published 2 months ago — by Nature
An AI model called DrugReflector, trained on gene expression data, significantly accelerates drug discovery by more effectively identifying promising compounds, doubling success rates when iteratively refined, and offering a powerful new tool for developing therapies.
Originally Published 3 months ago — by Nature
A new trimodal protein language model enhances the accuracy and efficiency of protein searches, leveraging advanced deep learning techniques to improve understanding of protein structures and functions.
Originally Published 4 months ago — by Quanta Magazine
The article discusses the resurgence of 'world models' in AI research, a concept dating back to the 1940s, which involves creating internal representations of the environment to improve AI decision-making and robustness. While early attempts relied on handcrafted models, modern deep learning approaches aim to develop these models automatically, though current systems often rely on heuristics rather than coherent representations. Developing effective world models is seen as crucial for advancing AI safety, reliability, and interpretability, with various approaches being explored to achieve this goal.
Originally Published 4 months ago — by Nature
BindCraft is a novel computational pipeline that leverages deep learning, specifically AlphaFold2, to de novo design functional protein binders targeting diverse proteins, including challenging membrane receptors, allergens, and nucleases, with high success rates and potential therapeutic applications.
Originally Published 4 months ago — by MIT News
A study by MIT researchers shows that simpler, physics-based models can outperform complex deep-learning models in climate prediction, especially for regional temperature estimates, highlighting the importance of appropriate benchmarking and problem-specific modeling approaches in climate science.
Originally Published 4 months ago — by Nature
AI tools like Evo, regLM, and AlphaGenome are advancing the decoding of the genome's regulatory regions, especially the non-coding 'dark matter' DNA, by predicting functions, interactions, and impacts of genetic variants, thus deepening our understanding of gene regulation and expression.
Originally Published 5 months ago — by SciTechDaily
Researchers have developed EchoNext, an AI tool that analyzes standard ECGs to detect hidden structural heart diseases more accurately than cardiologists, potentially transforming routine cardiac tests into early screening methods for serious heart conditions.
Originally Published 7 months ago — by TechCrunch
This article provides a comprehensive glossary of key AI terms, explaining concepts like AGI, AI agents, chain of thought, deep learning, diffusion, distillation, fine-tuning, GANs, hallucinations, inference, large language models, neural networks, training, transfer learning, and weights, to help readers understand the complex language used in AI research and development.
Originally Published 1 year ago — by MIT News
MIT researchers have developed a new photonic chip that performs deep neural network computations using light, achieving ultrafast processing speeds and high energy efficiency. This fully integrated photonic processor can handle both linear and nonlinear operations on-chip, overcoming previous limitations that required off-chip electronics. The chip's performance is comparable to traditional hardware, completing tasks in less than half a nanosecond with over 92% accuracy. This advancement could lead to faster, more efficient AI applications in fields like telecommunications and scientific research.
Originally Published 1 year ago — by Tech Explorist
Researchers at Washington State University have developed an AI system that uses deep learning to identify diseases from tissue images more accurately and quickly than humans. This model, detailed in Scientific Reports, processes gigapixel histopathology slides, significantly speeding up disease research. It integrates AI, computer vision, and medicine to overcome challenges in automatic disease detection, and has shown superior performance in identifying pathologies compared to previous systems and human experts.
Originally Published 1 year ago — by Neuroscience News
A deep learning AI model developed by Washington State University researchers significantly enhances the speed and accuracy of disease detection in tissue images, outperforming human pathologists in some cases. By analyzing gigapixel images with advanced neural networks, the AI reduces analysis times from months to weeks, revolutionizing research and diagnostics, particularly for cancer and gene-related illnesses. This model, already aiding animal disease research, holds transformative potential for human medical diagnostics.
Originally Published 1 year ago — by Phys.org
Scientists at Los Alamos National Laboratory have developed a new AI model, EPBDxDNABERT-2, which enhances the prediction of transcription factor binding to DNA, a key process in gene regulation related to diseases like cancer. By integrating DNA breathing dynamics into a deep learning framework, the model significantly improves the accuracy of predicting where transcription factors bind on the genome, aiding potential drug development. The model was trained using extensive gene sequencing data and tested on the Venado supercomputer, showing a 9.6% improvement in prediction accuracy.
Originally Published 1 year ago — by Ars Technica
The deep learning boom was inadvertently launched by computer scientist Fei-Fei Li, who created the ImageNet dataset, a massive collection of labeled images that proved crucial for training neural networks. Despite skepticism, Li's work, combined with advancements in GPU technology by Nvidia and the persistence of researchers like Geoffrey Hinton, enabled breakthroughs in AI, exemplified by the success of the AlexNet model in 2012. This convergence of big data, neural networks, and GPU computing marked a pivotal moment in AI history.