Valentina Pedoia

Title: Valentina Pedoia: Inventor in Machine Learning and Medical Imaging.

Introduction

Valentina Pedoia is an inventor based in San Francisco, California. Her patent work includes a machine-learning technique for semantic segmentation of sparse, multidimensional data.

Latest Patents

Pedoia is associated with one patent, titled “Semantic Segmentation of Sparse Multi-Dimensional Tensors.” The invention describes training a machine-learning model to perform semantic segmentation by iteratively extracting 3D image patches at two different sizes. The model is trained on the larger patches and their associated losses, followed by the smaller patches and their associated losses. This approach provides a method for training on image data at multiple spatial scales.

Career Highlights

Pedoia is associated with Altos Labs, Inc. Her listed patent work focuses on machine learning and the analysis of three-dimensional image data. She is credited with 1 patent.

Collaborations

Pedoia’s listed coworkers include Wilmer Arbelo-Gonzalez and Adam Scott Frost. The patent’s described approach reflects collaborative work in machine learning and image analysis.

Conclusion

Valentina Pedoia’s patent contribution centers on a multiscale training technique for semantic segmentation of 3D image data, connecting machine learning with image-analysis innovation.