Celeba
PulseAugur coverage of Celeba — every cluster mentioning Celeba across labs, papers, and developer communities, ranked by signal.
7 day(s) with sentiment data
-
Trigger color significantly impacts federated learning backdoor attack success
Researchers have demonstrated that the color of visual triggers significantly impacts the success rate of backdoor attacks in federated learning. By manipulating trigger colors on semantic objects like masks and sunglas…
-
New theory explains flow-based solvers, proposes efficient sampling method
Researchers have developed a new theoretical framework for understanding flow-based inverse solvers, which are used to solve imaging inverse problems. The new approach, termed posterior-transport, reveals that condition…
-
Flow Map Denoisers offer continuous control over image restoration tradeoffs
Researchers have introduced a novel method called Flow Map Denoisers, which addresses the fundamental tradeoff in image restoration between minimizing error and maximizing perceptual quality. This new approach utilizes …
-
InfantFace model enhances neonatal face detection in clinical settings
Researchers have developed InfantFace, a specialized face detection model based on the YOLOv11m architecture, designed for use in neonatal clinical environments. The model addresses challenges like cluttered backgrounds…
-
Lightweight U-Net uses YOLO-World heatmaps for face super-resolution
Researchers have developed a lightweight U-Net architecture for face super-resolution, capable of reconstructing high-resolution images from severely degraded inputs with an 8x magnification. A novel approach uses heatm…
-
polyDAG framework improves causal discovery in visual graphs
Researchers have developed polyDAG, a new framework for efficiently discovering causal relationships in visual semantic graphs. This method replaces computationally expensive acyclicity constraints with a polynomial tra…
-
Flow model optimizes compressed sensing for image reconstruction
Researchers have developed a novel flow-based generative model designed to optimize sampling policies in compressed sensing applications. This framework, which adapts the Flow Matching training paradigm, learns to selec…
-
New Method Identifies and Mitigates Bias in Vision Models Without Retraining
Researchers have developed a novel post-hoc method to identify and mitigate bias in frozen vision models without requiring additional labels or retraining. The technique uses gradient probes on concept decompositions to…
-
New Diffusion-Guided Smoothing Enhances Counterfactual Distribution Learning
Researchers have developed new diffusion-guided estimators for counterfactual distribution learning in high-dimensional settings. These methods employ geometry-adaptive localization, driven by diffusion score informatio…
-
Lie Group VAEs tackle non-commutative latent space challenges
Researchers have developed a new framework for Variational Autoencoders (VAEs) called Lie Group VAEs to better handle non-commutative structures in latent spaces. Traditional VAEs often enforce commutativity, which can …
-
Entropic Autoencoders Mitigate VAE Posterior Collapse
Researchers have introduced Entropic Autoencoders (EAEs), a novel framework designed to overcome the posterior collapse issue inherent in traditional Variational Autoencoders (VAEs). EAEs implicitly generate latent vari…
-
Optical networks achieve superior image denoising via pre-training
Researchers have developed a novel pre-training method for all-optical image denoising using diffractive networks. This approach involves an initial training phase with a large dataset of 3.45 million images, followed b…
-
New STMD method speeds diffusion model inference without teacher
Researchers have developed Stochastic Transition-Map Distillation (STMD), a novel framework designed to accelerate the inference process for diffusion models without requiring a pre-trained teacher model. This method di…
-
New framework enhances privacy in federated learning for sensitive data
Researchers have developed a new framework called the Gaussian Privacy Protector (GPP) designed to enhance privacy in data release, particularly for continuous, high-dimensional inputs. GPP utilizes a stochastic encoder…
-
ProtoFair introduces fair self-supervised learning by using pseudo-counterfactual pairs
Researchers have introduced ProtoFair, a novel method for enhancing fairness in self-supervised learning models. This approach integrates with existing self-supervised learning frameworks without requiring modifications…
-
New AI methods enhance out-of-distribution detection and representation learning
Researchers have developed UFCOD, a novel framework for few-shot cross-domain out-of-distribution (OOD) detection. UFCOD leverages information-geometric analysis of diffusion trajectories, extracting 'Path Energy' and '…