Vinith Kishore et al.: ICECREAM: high-fidelity equivariant cryo-electron tomography ##CryogenicElectronTomography##SelfSupervisedLearning##MachineLearning@insadelyon...##IUCrhttps://journals.iucr.org/paper?S2059798326001622
ICECREAM: high-fidelity equivariant cryo-electron tomography
We present ICECREAM, a self-supervised approach that achieves substantially better denoising and more reliable missing-wedge recovery in cryo-ET, while reducing training and inference time relative to comparable baselines.
journals.iucr.org
March 28, 2026 at 1:00 AM
Vinith Kishore et al.: ICECREAM: high-fidelity equivariant cryo-electron tomography ##CryogenicElectronTomography##SelfSupervisedLearning##MachineLearning@insadelyon...##IUCrhttps://journals.iucr.org/paper?S2059798326001622
ICECREAM bridges earlier self-supervised methods with the recent equivariant imaging framework for cryo-ET reconstruction #CryogenicElectronTomography #SelfSupervisedLearning #MachineLearning doi.org/10.1107/S205...
May 26, 2026 at 1:45 PM
ICECREAM bridges earlier self-supervised methods with the recent equivariant imaging framework for cryo-ET reconstruction #CryogenicElectronTomography #SelfSupervisedLearning #MachineLearning doi.org/10.1107/S205...
Our self-supervised method using variational autoencoder automatically identified a cluster containing pathological HFOs based on morphologies.
doi.org/10.1111/epi....
#epilepsy #Epilepsia #ILAE #artificialintelligence #highfrequencyoscillation #machinelearning #selfsupervisedlearning
doi.org/10.1111/epi....
#epilepsy #Epilepsia #ILAE #artificialintelligence #highfrequencyoscillation #machinelearning #selfsupervisedlearning
July 31, 2025 at 5:05 PM
Our self-supervised method using variational autoencoder automatically identified a cluster containing pathological HFOs based on morphologies.
doi.org/10.1111/epi....
#epilepsy #Epilepsia #ILAE #artificialintelligence #highfrequencyoscillation #machinelearning #selfsupervisedlearning
doi.org/10.1111/epi....
#epilepsy #Epilepsia #ILAE #artificialintelligence #highfrequencyoscillation #machinelearning #selfsupervisedlearning
🎉 New paper: SSMB
A self-supervised keypoint detector that works directly on motion-blurred images
No deblurring preprocessing, no handcrafted detector, no external pseudo-labels
📄 Paper: arxiv.org/abs/2608.27181
#ComputerVision #KeypointDetection #SelfSupervisedLearning #ImageMatching
A self-supervised keypoint detector that works directly on motion-blurred images
No deblurring preprocessing, no handcrafted detector, no external pseudo-labels
📄 Paper: arxiv.org/abs/2608.27181
#ComputerVision #KeypointDetection #SelfSupervisedLearning #ImageMatching
August 31, 2026 at 4:27 AM
🎉 New paper: SSMB
A self-supervised keypoint detector that works directly on motion-blurred images
No deblurring preprocessing, no handcrafted detector, no external pseudo-labels
📄 Paper: arxiv.org/abs/2608.27181
#ComputerVision #KeypointDetection #SelfSupervisedLearning #ImageMatching
A self-supervised keypoint detector that works directly on motion-blurred images
No deblurring preprocessing, no handcrafted detector, no external pseudo-labels
📄 Paper: arxiv.org/abs/2608.27181
#ComputerVision #KeypointDetection #SelfSupervisedLearning #ImageMatching
A fascinating result from the new LeJEPA identifiability work:
the model can recover the true hidden variables of the world from raw observations — but only if those latent variables are Gaussian.
Beautiful theorem. Slightly terrifying implication.
bit.ly/3S7iKkI
#WorldModels #SelfSupervisedLearning
the model can recover the true hidden variables of the world from raw observations — but only if those latent variables are Gaussian.
Beautiful theorem. Slightly terrifying implication.
bit.ly/3S7iKkI
#WorldModels #SelfSupervisedLearning
LeJEPA, World Models, and the Strange Tyranny of the Gaussian
A new paper argues that LeJEPA can recover the hidden variables of the world from pixels, but only under one very specific condition: the…
bit.ly
May 28, 2026 at 11:23 AM
A fascinating result from the new LeJEPA identifiability work:
the model can recover the true hidden variables of the world from raw observations — but only if those latent variables are Gaussian.
Beautiful theorem. Slightly terrifying implication.
bit.ly/3S7iKkI
#WorldModels #SelfSupervisedLearning
the model can recover the true hidden variables of the world from raw observations — but only if those latent variables are Gaussian.
Beautiful theorem. Slightly terrifying implication.
bit.ly/3S7iKkI
#WorldModels #SelfSupervisedLearning
DINOv3: la IA de visión universal sin anotaciones
Un nuevo estándar en visión por computadora
🔗 youtube.com/watch?v=yNyM...
#IA #ComputerVision #SelfSupervisedLearning #DINOv3 #AIResearch #DeepLearning
Un nuevo estándar en visión por computadora
🔗 youtube.com/watch?v=yNyM...
#IA #ComputerVision #SelfSupervisedLearning #DINOv3 #AIResearch #DeepLearning
DINOv3: La IA de visión universal que supera límites sin anotaciones
YouTube video by En la mente de la máquina, Inteligencia Artificial
youtube.com
August 16, 2025 at 2:31 PM
DINOv3: la IA de visión universal sin anotaciones
Un nuevo estándar en visión por computadora
🔗 youtube.com/watch?v=yNyM...
#IA #ComputerVision #SelfSupervisedLearning #DINOv3 #AIResearch #DeepLearning
Un nuevo estándar en visión por computadora
🔗 youtube.com/watch?v=yNyM...
#IA #ComputerVision #SelfSupervisedLearning #DINOv3 #AIResearch #DeepLearning
Big congratulations to all authors! 🚀
#ICLR2026 #MachineLearning #AIResearch #RepresentationLearning #InformationRetrieval #DenseRetrieval #SelfSupervisedLearning #LanguageModels #NLP #UKPLab #ICLR2026
@cmu.edu @tencent.bsky.social @tuda.bsky.social @cs-tudarmstadt.bsky.social @microsoft.com
#ICLR2026 #MachineLearning #AIResearch #RepresentationLearning #InformationRetrieval #DenseRetrieval #SelfSupervisedLearning #LanguageModels #NLP #UKPLab #ICLR2026
@cmu.edu @tencent.bsky.social @tuda.bsky.social @cs-tudarmstadt.bsky.social @microsoft.com
February 4, 2026 at 11:40 AM
AstroM3 Transforms Astronomy With Multimodal AI for Stellar Insights 🌌🔭✨ www.azoai.com/news/2024111... #Astronomy #AI #MachineLearning #AstroTech #DataScience #VariableStars #AstroInnovation #SelfSupervisedLearning #Scientific @ucberkeleyofficial.bsky.social @berkeleylab.bsky.social
AstroM3 Transforms Astronomy With Multimodal AI for Stellar Insights
AstroM3, a self-supervised multimodal framework, integrates photometry, spectra, and metadata to transform astronomical data analysis. It improves classification, detects anomalies, and uncovers hidde...
www.azoai.com
November 20, 2024 at 12:09 AM
AstroM3 Transforms Astronomy With Multimodal AI for Stellar Insights 🌌🔭✨ www.azoai.com/news/2024111... #Astronomy #AI #MachineLearning #AstroTech #DataScience #VariableStars #AstroInnovation #SelfSupervisedLearning #Scientific @ucberkeleyofficial.bsky.social @berkeleylab.bsky.social
Next-Embedding Prediction Makes Strong Vision Learners
Joyce Chai, Saining Xie et al.
Paper
Details
#SelfSupervisedLearning #VisionModels #DeepLearning
Joyce Chai, Saining Xie et al.
Paper
Details
#SelfSupervisedLearning #VisionModels #DeepLearning
December 19, 2025 at 9:02 AM
Next-Embedding Prediction Makes Strong Vision Learners
Joyce Chai, Saining Xie et al.
Paper
Details
#SelfSupervisedLearning #VisionModels #DeepLearning
Joyce Chai, Saining Xie et al.
Paper
Details
#SelfSupervisedLearning #VisionModels #DeepLearning
I discovered this thought-provoking paper about using masked boundary modeling to help vision models learn dense spatial details from boundary cues, outperforming DINOv3 and improving depth for embodied AI. See link below. #ComputerVision #SelfSupervisedLearning.
https://arxiv.org/abs/2607.05247
https://arxiv.org/abs/2607.05247
July 8, 2026 at 1:19 PM
I discovered this thought-provoking paper about using masked boundary modeling to help vision models learn dense spatial details from boundary cues, outperforming DINOv3 and improving depth for embodied AI. See link below. #ComputerVision #SelfSupervisedLearning.
https://arxiv.org/abs/2607.05247
https://arxiv.org/abs/2607.05247
At the #ML4ITS2023 Workshop at @ecmlpkdd.org, Damien describes how he uses language models and #GPT architecture to obtain a contextualized representation of Activities of Daily Living #ADL for classification and transfer learning
👉 inria.hal.science/hal-04858744v1
#LLM #selfSupervisedLearning
👉 inria.hal.science/hal-04858744v1
#LLM #selfSupervisedLearning
May 26, 2025 at 9:13 AM
At the #ML4ITS2023 Workshop at @ecmlpkdd.org, Damien describes how he uses language models and #GPT architecture to obtain a contextualized representation of Activities of Daily Living #ADL for classification and transfer learning
👉 inria.hal.science/hal-04858744v1
#LLM #selfSupervisedLearning
👉 inria.hal.science/hal-04858744v1
#LLM #selfSupervisedLearning
👉 Let’s talk multimodal learning — come by or check out the paper!
Joint work with an amazing team: @benoit-dufumier.bsky.social, D. Tuia & J-P. Thiran
#ICLR2025 #MultimodalLearning #SelfSupervisedLearning #ContrastiveLearning
Joint work with an amazing team: @benoit-dufumier.bsky.social, D. Tuia & J-P. Thiran
#ICLR2025 #MultimodalLearning #SelfSupervisedLearning #ContrastiveLearning
April 24, 2025 at 4:02 PM
👉 Let’s talk multimodal learning — come by or check out the paper!
Joint work with an amazing team: @benoit-dufumier.bsky.social, D. Tuia & J-P. Thiran
#ICLR2025 #MultimodalLearning #SelfSupervisedLearning #ContrastiveLearning
Joint work with an amazing team: @benoit-dufumier.bsky.social, D. Tuia & J-P. Thiran
#ICLR2025 #MultimodalLearning #SelfSupervisedLearning #ContrastiveLearning
The A²SL framework retrieves relevant observations to boost predictions of water temperature and dissolved oxygen, showing accuracy in sparse and extreme conditions. https://getnews.me/adaptive-self-supervised-learning-enhances-environmental-modeling/ #selfsupervisedlearning #environmentalmodeling
September 19, 2025 at 8:18 PM
The A²SL framework retrieves relevant observations to boost predictions of water temperature and dissolved oxygen, showing accuracy in sparse and extreme conditions. https://getnews.me/adaptive-self-supervised-learning-enhances-environmental-modeling/ #selfsupervisedlearning #environmentalmodeling
A spatiotemporal self‑supervised model trained on 3,161 MRI scans outperformed supervised learning on six of seven Alzheimer’s prediction tasks; the code is publicly released. https://getnews.me/self-supervised-learning-improves-mri-prediction-of-alzheimers/ #selfsupervisedlearning #alzheimers
September 17, 2025 at 7:16 AM
A spatiotemporal self‑supervised model trained on 3,161 MRI scans outperformed supervised learning on six of seven Alzheimer’s prediction tasks; the code is publicly released. https://getnews.me/self-supervised-learning-improves-mri-prediction-of-alzheimers/ #selfsupervisedlearning #alzheimers
Published by: Yiming Lei, Junping Zhang, Hongming Shan📅 Published on: October 26, 2021📄 Full Study: [DOI: 10.1007/s43657-021-00025-y] 🔗#MedicalAI #CTDenoising #LungNoduleClassification #SelfSupervisedLearning #Bioinformatics
April 1, 2026 at 3:16 AM
Published by: Yiming Lei, Junping Zhang, Hongming Shan📅 Published on: October 26, 2021📄 Full Study: [DOI: 10.1007/s43657-021-00025-y] 🔗#MedicalAI #CTDenoising #LungNoduleClassification #SelfSupervisedLearning #Bioinformatics
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
AI Is Learning How to Perfect Lighting in Photos
hackernoon.com
December 21, 2024 at 4:01 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Beeble Researchers Develop AI That Can Make Any Photo Look Perfectly Lit—Even in the Darkest Room
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers Build Massive AI Training Dataset to Perfect Lighting on Faces
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers Develop AI That Can Copy, Adjust, and Perfect Lighting in Any Image
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers Asked 47 People to Judge AI-Enhanced Portraits—Here’s What They Chose
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Beeble Researchers Attempt to Manipulate Lighting on Digital Image Using AI
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
This New Tech Makes Faces in Photos Look Incredibly Realistic
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
AI Learns to Perfect Lighting in Photos Using Smart Masks and Creative Training
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Scientists Use Portraits and Balls to Test an AI’s Relighting Abilities
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning
AI System Uses Teamwork to Create Picture-Perfect Lighting
hackernoon.com
December 21, 2024 at 4:00 PM
Researchers at Beeble AI have developed a method for improving how light and shadows can be applied to human portraits in digital images. #selfsupervisedlearning