K
Kevin Smith
Researcher at Royal Institute of Technology
Publications - 60
Citations - 10918
Kevin Smith is an academic researcher from Royal Institute of Technology. The author has contributed to research in topics: Image segmentation & Breast cancer. The author has an hindex of 25, co-authored 60 publications receiving 8802 citations. Previous affiliations of Kevin Smith include Karolinska Institutet & École Polytechnique Fédérale de Lausanne.
Papers
More filters
Book ChapterDOI
Decoupling Inherent Risk and Early Cancer Signs in Image-Based Breast Cancer Risk Models
TL;DR: This work trains networks using three different criteria to select the positive training data, and finds that these three models learn distinctive features that focus on different patterns, which translates to contrasts in performance.
Book ChapterDOI
Multi-person tracking in meetings: a comparative study
TL;DR: The focus of the study was to test and evaluate various multi-person tracking methods developed in theAMI project using a standardized data set and evaluation methodology.
Proceedings Article
The Preimage of Rectifier Network Activities
TL;DR: A procedure for explicitly computing the complete preimage of activities of a layer in a rectifier network with fully connected layers, from knowledge of the weights in the network, is given.
Proceedings ArticleDOI
Automated quantification of morphodynamics for high-throughput live cell time-lapse datasets
TL;DR: This work presents a fully automatic method to track and quantify the morphodynamics of differentiating neurons in fluorescence time-lapse datasets and confirms with strong statistical significance static and dynamic behaviors that had been previously observed by biologists, but never measured.
Posted Content
Adding Seemingly Uninformative Labels Helps in Low Data Regimes
Christos Matsoukas,Albert Bou I Hernandez,Yue Liu,Karin Dembrower,Gisele Helena Barboni Miranda,Emir Konuk,Johan Fredin Haslum,Athanasios Zouzos,Peter Lindholm,Fredrik Strand,Kevin Smith +10 more
TL;DR: This work considers a task that requires difficult-to-obtain expert annotations: tumor segmentation in mammography images, and shows that performance can be improved by complementing the expert annotations with seemingly uninformative labels from non-expert annotators, turning the task into a multi-class problem.