B-spline
PulseAugur coverage of B-spline — every cluster mentioning B-spline across labs, papers, and developer communities, ranked by signal.
5 day(s) with sentiment data
-
PyTorch library TorchCurves introduces learnable parametric curves
A new PyTorch library called TorchCurves has been introduced, enabling the use of vector-valued parametric curves as learnable components within machine learning models. This library allows for the optimization of curve…
-
New FIND framework enhances LiDAR 3D detection by disentangling frequency domains
Researchers have introduced FIND (Frequency Invariance Disentanglement), a new framework designed to improve LiDAR-based 3D Object Detection (L3OD) performance by addressing domain shifts. FIND operates in the frequency…
-
New method generates closed-form space-time covariance functions
A new paper introduces a method for creating closed-form covariance functions in space and time using B-spline spectral bases. This approach, derived from Bochner's Theorem, allows for non-parametric descriptions of aut…
-
Kolmogorov-Arnold Networks: Benefits questioned, educational tool released
A new paper questions the universal benefits of Kolmogorov-Arnold Networks (KANs), suggesting their favorable scaling properties may be tied to specific target structures rather than an inherent architectural advantage.…
-
KANLab offers interactive browser tool for learning Kolmogorov-Arnold Networks
KANLab is a new open-source, browser-based tool designed to help users learn about Kolmogorov-Arnold Networks (KANs). It visualizes B-spline activation curves, gradient descent, and pruning in a simplified 2D environmen…
-
LLMs enhance B-spline surface fitting in CAD systems
Researchers have developed a new framework called LASP that uses large language models (LLMs) to improve B-spline surface fitting in computer-aided design (CAD) systems. This framework translates modeling histories into…
-
New T-KAN architecture boosts high-frequency trading forecasts
Researchers have introduced Temporal Kolmogorov-Arnold Networks (T-KAN), a novel architecture designed to improve high-frequency trading (HFT) forecasting. Unlike traditional models that struggle with noisy, non-linear …
-
New FAN-LoRA method improves medical image segmentation for foundation models
Researchers have developed FAN-LoRA, a new method for adapting vision foundation models like the Segment Anything Model (SAM) to medical imaging domains. Existing methods struggle with domain shifts, leading to performa…
-
New mesh-free neural network models brittle fracture with high accuracy
Researchers have developed a novel mesh-free method using a single neural network to model brittle fracture, eliminating the need for explicit crack tracking. This approach employs a multiresolution feature encoding bas…
-
New B-Spline Framework Enhances 3D Tooth Segmentation Accuracy
Researchers have developed a novel framework called B-Spline Embedded Structure Learning for accurate 3D tooth segmentation. This method addresses challenges in digital dentistry posed by complex dental arrangements lik…
-
New research contrasts B-Spline and neural implicit methods for image registration
A new research paper explores the impact of deformation priors in continuous deformable image registration methods. The study compares traditional B-Spline transformations with newer neural implicit representation (INR)…
-
New method reconstructs 3D geometry for direct CAD and simulation use
Researchers have developed FORGE-SIM, a novel method for reconstructing 3D geometry from sparse RGB images that is directly compatible with CAD and simulation workflows. This approach optimizes a multi-patch B-spline bo…
-
New PI-Splines Architecture Offers Stable Alternative for Physics-Informed Learning
Researchers have introduced Physics-Informed Splines (PI-Splines), a novel architecture for physics-informed learning that directly parametrizes unknown fields using B-spline expansions. This method offers advantages ov…
-
New PIE-PINN framework enhances elastic property estimation from noisy data
Researchers have developed a Probabilistic Inverse Elasticity Physics-Informed Neural Network (PIE-PINN) framework designed to robustly estimate heterogeneous elastic properties from noisy and low-resolution displacemen…
-
New research details 3D point cloud attribute compression via deep unrolling
A new research paper on arXiv introduces a method for compressing attributes of 3D point clouds. The approach utilizes a multi-resolution B-spline framework and a feed-forward network derived from rate-distortion optimi…
-
DynFly framework enhances UAV navigation with continuous trajectory generation
Researchers have developed DynFly, a new framework designed to improve the continuous trajectory generation for unmanned aerial vehicles (UAVs) in urban environments. This system addresses the gap between high-level nav…
-
New framework enhances medical image super-resolution with dual-prior learning
Researchers have developed a new framework called Dual-Prior Null-space Learning (DP-NSL) for arbitrary slice super-resolution in medical imaging. This method reconstructs isotropic volumes from anisotropic clinical acq…
-
New metrics assess hardware inference complexity of Kolmogorov-Arnold Networks
A new paper introduces hardware-oriented metrics for evaluating the inference complexity of Kolmogorov-Arnold Networks (KANs). These metrics, including Real Multiplications (RM), Bit Operations (BOP), and Number of Addi…
-
New KAN Frameworks and Variants Enhance Research and Efficiency
Researchers have developed KANLib, a new framework designed to streamline research on Kolmogorov-Arnold Networks (KANs) by unifying features from existing implementations like PyKAN, EfficientKAN, and FastKAN. Concurren…
-
New 3D Trace World Model Enhances Scalable Robot Learning
Researchers have developed $\mu_0$, a novel world model for robotics that utilizes 3D interaction traces to predict the movement of salient objects and points. This approach bypasses the need for embodiment-specific act…