Lmst

Organized with @DrVeronikaCH , Théo Sourget and Steff Groefsema. Website: https://medical-datasets.github.io/webinar/

The webinar will be hosted on Zoom, register here https://itucph.zoom.us/meeting/register/-koafnhwR9aR3uhBEPcfag

Looking forward to your participation in making this a great event! 😊

#MakingMetadataCount #FAccT #FAccT2025

#MakingMetadataCount #FAccT2025 #FAccT

👩‍💻 If you're interested in learning more, I'll be presenting this work in our webinar series on May 12th at 10:00 CEST. Join us here https://medical-datasets.github.io/webinar/

🐈‍⬛ Copycats has been accepted at #NeurIPS2024 D&B Track! 🎉🎉

TL; DR: We aim to promote better data governance in the context of medical imaging datasets and uphold the reliability and fairness of AI in healthcare.

🔗 https://arxiv.org/abs/2402.06353

#MakingMetadataCount with contributions of great collaborators 🤩

A big thank you and congratulations Natalia-Rozalia Avlona, Dovile Juodelyte, Théo Sourget, Caroline Vang-Larsen, Anna Rogers, Hubert Dariusz Zając, @DrVeronikaCH 🤗

This slide shows the differences between medical imaging and computer vision: de-identification for patient data, images from different patients, metadata such as patient demographics or hospital scanners. Treating medical imaging as "small computer vision": images + labels but ignoring metadata can lead to inaccurate bias results.

We provide some recommendations: 1) Access to datasets should be predictable, open-licensing and persistent. 2) Evaluation should be performed against the context of real-world use. 3) Documentation should be complete and up-to-date. 4) Community-Contributed Platforms could benefit from commons-based governance.

We aim to promote better data governance in the context of medical imaging datasets and uphold the reliability and fairness of AI models in healthcare.

“Datasets through the Looking Glass” has been featured in RSIP Vision! 😊 Check out our last webinar on "dataset design and metrics" with Dr. Annika Reinke, Natalia-Rozalia Avlona, Hubert Dariusz Zając, Dr. Sandra Avila and Alceu Bissoto

📖 Article: https://rsipvision.com/ComputerVisionNews-2024April//50/

🎥 https://rb.gy/2pmcmb

#MakingMetaDataCount with
@DrVeronikaCH

🔗 Webinar website: https://purrlab.github.io/webinar

📫 Stay updated, sign up to our newsletter: https://tinyurl.com/27a9syb7

Datasets through the Looking-Glass is a webinar series focused on reflecting on data-related facets of machine learning. The webinar is part of the "Making Metadata Count" project, organized by Veronika Cheplygina and Amelia Jiménez Sánchez (both on the picture). This webinar series was about dataset design and ML evaluation, we had 3 talks with 5 speakers. Hubert Dariusz Zając and Natalia-Rozalia Avlona were the speakers of the first talk. Their research is a critical inspection of ground truth labels when working with medical imaging datasets.

Dr. Annika Reinke was the speaker of the second talk. She presented "Metrics Reloaded", a problem-aware metric recommendation framework resulting from 2.5 years of research involving a total of 73 international experts.

The speakers of our last talk were Alceu Bissoto and Dr. Sandra Avila. First, Sandra presented an overview of their group work on skin lesions over the last 10 years, covering areas such as data augmentation, transfer learning and generative AI. At this point, they found that their skin lesions were so good that they needed to start thinking about bias. Then, Alceu presented a strategy to evaluate bias on skin lesion models, in particular “trap sets”.

As a final note, good luck to Hubert and Alceu in their upcoming PhD defenses :)

🎉 Exiting news! Join us for our next "Datasets through the L👀king Glass" webinar featuring amazing speakers 🤩

🗣️ Dataset design and metrics
📅 25th Mar 2024 - 4pm CET
🖥️ Zoom https://rb.gy/jfx71z
🔗 https://purrlab.github.io/webinar/

#MakingMetaDataCount with
@DrVeronikaCH

📫 If you want to receive information about upcoming seminars, sign up to our mailing list! https://tinyurl.com/27a9syb7

Webinar title: Datasets through the Looking-Glass. Topic: Dataset design and metrics. 25th of March 2024 at 4pm CET on Zoom. Speakers: Dr. Annika Reinke, Hubert Dariusz Zając, Natalia-Rozalia Avlona, Alceu Bissoto, Dr. Sandra Avila. Organized by Amelia Jiménez-Sánchez and Veronika Cheplygina at IT University of Copenhagen. Website: https://purrlab.github.io

“Datasets through the L👀king Glass” has been featured in RSIP Vision magazine! Check out our last webinar on "shortcuts and bias" ☺️ thanks!

#MakingMetaDataCount with @DrVeronikaCH

📖 Read the full article: https://www.rsipvision.com/ComputerVisionNews-2024January/

🎥 All previous webinar talks are available on our YouTube playlist https://rb.gy/2pmcmb

Picture on the top left showing Veronika (left) and Amelia (right). Description about the topic and goal of the webinar, and summary of the biographies and talks of our speakers. Our speakers for this webinar on "shortcuts and bias" were Jessica Schrouff (Google DeepMind), Rhys Compton and Lily Zhang (New York University), and Enzo Ferrante (Argentina’s National Research Council). The videos of their talks are embedded (Youtube link).

Our webinar "Datasets through the L👀king-Glass" is back 😎 with great speakers: Dr. Jessica Schrouff, Rhys Compton, Lily Zhang and Dr. Enzo Ferrante.

🗣️ Shortcuts and bias
📅 4th Dec 2023 - 4 pm CET
🖥️ Zoom https://shorturl.at/cetY3
🔗 https://purrlab.github.io/webinar/

#MakingMetaDataCount with @DrVeronikaCH

Title: Datasets through the Looking Glass, subtitle: Shortcuts and bias. Webinar is on 4th December at 4pm CET in Zoom. Speakers are Dr. Jessica Schrouff (Google DeepMind, UK), Rhys Compton and Lily Zhang (New York University, USA), Dr. Enzo Ferrante (CONICET, Argentina). Organized by Amelia Jiménez-Sánchez and Veronika Cheplygina.

Datasets through the Looking-Glass is a webinar series focusing on the data aspects on learning-based methods. Our aim is to build a community of scientists interested in understanding how the data we use affects the algorithms and society instead of only optimizing for a performance metric.

Next webinar on 4th December 2023 at 4pm CET, topic: shortcuts and bias.

Speakers: Dr. Jessica Schrouff (Google DeepMind, UK), Dr. Enzo Ferrante (CONICET, Argentina), Rhys Compton and Lily Zhang (New York University, USA).

Organizers at IT University of Copenhagen: Dr. Amelia Jiménez Sánchez and Dr. Veronika Cheplygina.

Our webinar “Datasets through the L👀king Glass” has been featured at RSIP Vision ☺️ thank you!

Don’t miss Andre Pacheco presenting PAD-UFES-20! They collected smartphone pictures to compose a skin lesion dataset with various metadata in Brazil.

#MakingMetaDataCount with @DrVeronikaCH

🎥 Video is available on our website: https://purrlab.github.io/webinar/

📖 Read the full article: https://www.rsipvision.com/ComputerVisionNews-2023October/44/

Picture of Veronika and Amelia. The webinar series “Datasets through the L👀king Glass” focuses on the data aspects of machine learning algorithms. It’s part of the project “Making MetaDataCount” at IT University of Copenhagen. We investigate shortcuts (image artifacts or demographics) in medical image datasets. In the last webinar, Andre Pacheco presented PAD-UFES-20, a skin lesion dataset composed of smartphone pictures and various meta-data.

We'll be starting 🔜

#MakingMetaDataCount with @DrVeronikaCH

https://social.itu.dk/@amelia/111069181063913145

‼️ Don't forget to register for our next webinar on Monday 18 Sept 👉 http://shorturl.at/nDKV1

🗣️ Cathrine Damgaard and Trine N. Eriksen, Dr. Thijs Kooi, and Dr. Andre Pacheco
📅 18 Sept 2023 - 1pm CEST

🔗 https://purrlab.github.io/webinar/

#MakingMetaDataCount with @DrVeronikaCH

📢 Webinar update: new confirmed speakers 😊

Our recently graduated MSc students: Cathrine Damgaard and Trine N. Eriksen 👩‍🎓

will join Dr. Thijs Kooi and Dr. Andre Pacheco at the next webinar!

🗣️ Rethinking Annotations
📅 18th Sept 2023 - 1pm CEST
💻 Zoom: https://t.co/NetLL6YwGF

🔗 http://purrlab.github.io/webinar/

#MakingMetaDataCount with
@DrVeronikaCH

Our webinar "Datasets through the L👀king-Glass" is back 😎 and we will have great speakers: Dr. Thijs Kooi and Dr. Andre Pacheco

🗣️ Rethinking Annotations
📅 18th Sept 2023 - 1pm CEST
🖥️ Zoom https://shorturl.at/nDKV1

#MakingMetaDataCount with @DrVeronikaCH

🔗 http://purrlab.github.io/webinar/

Many people working on #MachineLearning algorithms assume the data is already there, but there's a lot of work behind the scenes! If you want to learn about how some #MedicalImaging challenge datasets were made, join our next "Datasets through the L👀king-Glass" webinar, organized by @amelia

🗓️ Date: 5th June 2023
🕓 4pm CEST
💻 Where: Zoom. Register at https://shorturl.at/horRW

🔗 https://purrlab.github.io/webinar/

#MakingMetadataCount

"Datasets through the L👀king-Glass" is back!

Next talk:
🗣️ Datasets for medical image challenges
🗓️ Date: 5th June 2023
🕓 4pm CEST
💻 Where: Online - Zoom. Register here https://shorturl.at/horRW

#MakingMetaDataCount with @DrVeronikaCH
🔗 https://purrlab.github.io/webinar/

The webinar series is called "Datasets through the Looking-Glass". The next webinar will be about "Datasets for medical image challenges" on 5th of June at 4pm CEST hosted on Zoom. The speakers will be Prof. Amber Simpson, Dr. Esther E. Bron, Prof. Spyridon Bakas and Dr. Ujjwal Baid.

Our webinar series "Datasets through the L👀king-Glass", will kick-off soon:
🗣️ Characteristics of skin lesions datasets
🗓️ Date: 27th Feb 2023
🕓 4pm CET
💻 Where: Online - Zoom http://shorturl.at/coS12

#MakingMetaDataCount with @DrVeronikaCH
🔗 https://purrlab.github.io/webinar/

#MakingMetadataCount

Client Info