Mae
PulseAugur coverage of Mae — every cluster mentioning Mae across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
New AnomNOVIC system enables robots to identify unknown objects without prompts
Researchers have developed AnomNOVIC, a novel framework for robots to recognize unseen objects in real-world environments. This system combines a masked autoencoder (MAE) for anomaly detection with the NOVIC image class…
-
New GMM pooling method enhances preterm birth prediction from ultrasound images
Researchers have developed a new Gaussian Mixture Model (GMM) pooling method for multiple instance learning (MIL) to improve preterm birth prediction from ultrasound images. This approach models the feature distribution…
-
CUPID deepfake detector uses UV maps and MAE for interpretable analysis
Researchers have developed CUPID, a novel deepfake detection method that reconstructs UV texture maps from 3D face models and utilizes Masked Autoencoders (MAE) for analysis. This approach does not require deepfake vide…
-
New method improves electricity load forecasting with deep learning
Researchers have developed a delta-based target reformulation method for short-term electricity load forecasting using deep learning models like LSTMs and Transformers. This approach predicts the change in load between …
-
New Drift-RAE Method Enhances Representation Autoencoder Distillation
Researchers have developed a new method called Drift-RAE to improve the distillation process for representation autoencoders (RAEs). This technique addresses issues of anisotropy and large curvatures in RAE latent space…
-
New MoFore Framework Advances Self-Supervised Video Representation Learning
Researchers have introduced MoFore, a novel framework for self-supervised video representation learning that focuses on forecasting future latent embeddings from distant context clips. Unlike previous methods that relie…
-
New RePAIR architecture learns chess concepts via self-supervised learning
Researchers have developed a new self-supervised learning architecture called RePAIR, which combines elements of MAE, JEPA, and BERT. This architecture is designed to encode sequential data, such as chess positions, int…
-
Self-Soupervision enables model soups from unlabeled data
Researchers have developed a new method called Self-Soupervision, which allows for the creation of "model soups" using self-supervised learning (SSL) instead of traditional supervised learning. This technique enables th…
-
New datasets and AI methods boost skin lesion classification
Researchers have developed new datasets and methods to improve AI's ability to classify skin lesions from dermatoscopic images. One paper introduces IMA++, a large dataset with over 17,000 segmentation masks from nearly…
-
New AI approach embraces forgetting for better domain adaptation
Researchers have developed a new method for domain incremental learning that embraces catastrophic forgetting rather than trying to prevent it. The approach uses domain-specific LoRA adapters and a self-supervised maske…
-
New LNN-PINN Framework Boosts Physics-Informed Neural Network Accuracy
Researchers have developed LNN-PINN, a new framework designed to enhance the accuracy of physics-informed neural networks (PINNs). This framework integrates a liquid residual gating architecture into the hidden layers o…
-
HEPA architecture predicts critical time-series events using self-supervision
Researchers have developed HEPA, a novel self-supervised architecture for predicting critical events in multivariate time series data. This architecture uses a causal Transformer encoder pretrained with a Joint-Embeddin…
-
New MVProbe framework analyzes AI models via weight-space learning
Researchers have developed MVProbe, a novel multi-view probing framework designed to analyze large open-source AI models directly from their parameters. This method addresses the computational limitations of processing …
-
Pretraining objective impacts low-data image classification
A new study on arXiv investigates the impact of different pretraining objectives on the performance of visual encoders in extreme low-data fine-grained classification tasks. Researchers compared four frozen ViT-B/16 enc…
-
AI framework enhances wearable health monitoring in harsh underwater conditions
Researchers have developed a memory-efficient framework for denoising electrodermal activity (EDA) signals, crucial for wearable health monitoring systems. The method employs knowledge distillation to train a lightweigh…
-
SSMProbe framework reveals importance of token order in visual representations
Researchers have developed SSMProbe, a new framework for analyzing visual representations in AI models. This method utilizes State Space Models (SSMs) to account for the critical role of token order, challenging the tra…
-
OpenAI-affiliated researchers integrate FID into training, achieving sub-0.8 ImageNet scores
Researchers from USC, CMU, CUHK, and OpenAI have developed a new method called FD-loss that allows the Fréchet Inception Distance (FID) metric to be directly incorporated into the training process of image generation mo…
-
AI model synthesizes liver MRI hepatobiliary phase images for better HCC detection
Researchers have developed a new deep learning model called the Triple-Phase Sequential Fusion Network (TriPF-Net) to synthesize hepatobiliary phase (HBP) liver MRI images. This network leverages sequential information …
-
Self-supervised MAE pretraining boosts nnFormer for medical image segmentation
Researchers have developed a self-supervised pretraining framework using Masked Autoencoders (MAE) to improve the efficiency of nnFormer models for medical image segmentation. This approach allows the model to learn ana…