MC Dropout
PulseAugur coverage of MC Dropout — every cluster mentioning MC Dropout across labs, papers, and developer communities, ranked by signal.
2 day(s) with sentiment data
-
New VBLL method enhances online node classification on evolving graphs
Researchers have developed a new method called variational Bayesian last-layer (VBLL) for online node classification on evolving graphs. This approach addresses the challenges of inductive generalization and calibrated …
-
New PI-CP method enhances uncertainty quantification for neural operators
Researchers have developed a new method called Physics-Informed Conformal Prediction (PI-CP) to provide reliable uncertainty estimates for neural operators used in approximating solutions to partial differential equatio…
-
Lévy Attention introduces single-pass predictive uncertainty for time series
Researchers have introduced Lévy Attention, a novel attention mechanism designed for irregularly sampled time series data. This new approach integrates predictive uncertainty directly into the attention layer, allowing …
-
AI paraphrasing improves sentiment classifier accuracy, study finds
A new study published on arXiv explores how sentiment classifiers perform on sarcastic and AI-paraphrased social media text. Researchers found that classifiers exhibit lower confidence scores on sarcastic content, indic…
-
GATTA framework enhances active learning with test-time augmentation
Researchers have introduced GATTA, a novel framework designed to improve active learning on graph-structured data by leveraging test-time augmentation (TTA). GATTA aggregates predictions from multiple augmented views to…
-
New AI framework enhances skin lesion classification with uncertainty and explainability
Researchers have developed a new framework for classifying skin lesions that combines deep ensemble learning with uncertainty quantification and explainability techniques. This approach uses multiple models, including v…
-
New research examines uncertainty in image segmentation models
A new paper on arXiv explores uncertainty quantification (UQ) in image segmentation, a critical area for safety-sensitive applications. The research investigates the interaction between aleatoric uncertainty (data-relat…
-
ST-LoRA: Parameter-Efficient Ensemble for Agricultural Segmentation
Researchers have developed ST-LoRA, a novel parameter-efficient ensemble framework designed for uncertainty-aware agricultural segmentation. This method combines Low-Rank Adaptation (LoRA) with snapshot ensembling to cr…
-
New hybrid deep learning model enhances RF modulation recognition
Researchers have developed a novel uncertainty-driven hybrid deep learning architecture for radio frequency (RF) modulation recognition. This system combines spectral information from FFT preprocessing with time-frequen…
-
New RL framework enhances trading confidence with uncertainty estimation
Researchers have developed a new framework for reinforcement learning (RL) in algorithmic trading that incorporates comprehensive uncertainty estimation. This approach addresses the challenges of dynamic financial marke…
-
Heckman correction improves ML model uncertainty calibration
Researchers have developed a new method for addressing epistemic uncertainty in machine learning models, particularly when training data is subject to selection bias. The proposed technique adapts the Heckman correction…
-
New deep learning model classifies astronomical transients without human labels
Researchers have developed a novel deep learning framework for classifying astronomical transients as real or bogus without requiring human-labeled data. This method utilizes injected simulated transients and a contamin…
-
Bayesian deep learning evaluation unstable in low-data settings, studies find
Two new arXiv papers highlight significant instability in evaluating Bayesian deep learning methods, particularly under data scarcity. Researchers found that standard evaluation metrics can produce unreliable and datase…