{ "id": 452360, "title": "UBC-OCEAN Challenge Presentation: Insights and Tips", "url": "/competitions/UBC-OCEAN/discussion/452360", "author_name": "masadia", "post_date": "2023-11-01T22:39:22.297000", "votes": 28, "comment_count": 29, "views": 0, "content": "

Dear participants,

\n

We're thrilled to share a presentation on the UBC-OCEAN challenge, featuring essential dataset information and valuable tips. The annotation files mentioned in slides 12 and 13 (time: 8:45 to 10:15) have been submitted to Kaggle and will be accessible to you in a few days.

\n

We remain dedicated to promptly providing the resources you need to enhance your model's performance.

\n

Thank you for your participation, and please stay tuned for updates.

\n

Best regards,
\nMaryam

", "messages": [ { "id": 2508705, "postDate": "2023-11-01T22:39:22.297Z", "content": "

Dear participants,

\n

We're thrilled to share a presentation on the UBC-OCEAN challenge, featuring essential dataset information and valuable tips. The annotation files mentioned in slides 12 and 13 (time: 8:45 to 10:15) have been submitted to Kaggle and will be accessible to you in a few days.

\n

We remain dedicated to promptly providing the resources you need to enhance your model's performance.

\n

Thank you for your participation, and please stay tuned for updates.

\n

Best regards,
\nMaryam

", "rawMarkdown": "Dear participants,\n\nWe're thrilled to share a presentation on the [UBC-OCEAN challenge](https://www.youtube.com/watch?v=49NP8VEOH5I), featuring essential dataset information and valuable tips. The annotation files mentioned in slides 12 and 13 (time: 8:45 to 10:15) have been submitted to Kaggle and will be accessible to you in a few days.\n\nWe remain dedicated to promptly providing the resources you need to enhance your model's performance.\n\nThank you for your participation, and please stay tuned for updates.\n\nBest regards,\nMaryam", "votes": 28 }, { "id": 2508771, "postDate": "2023-11-02T02:09:02.977Z", "content": "

In general, it is not recommended to add a different format of data (tumor non-tumor mask in this case?) in the middle of a competition.

\n

This is because the data may require us to change our approach significantly, which can be very burdensome.

\n

I don't think this was the case in the PANDA competition.

", "rawMarkdown": "In general, it is not recommended to add a different format of data (tumor non-tumor mask in this case?) in the middle of a competition. \n\nThis is because the data may require us to change our approach significantly, which can be very burdensome.\n\nI don't think this was the case in the PANDA competition.", "votes": 17, "replies": [ { "id": 2513776, "postDate": "2023-11-05T17:36:08.810Z", "rawMarkdown": "", "isDeleted": true }, { "id": 2517052, "postDate": "2023-11-08T07:16:19.320Z", "content": "

I think we have to restart our work after the annotation mask is available. But the left time before deadline is less than 2 months, and the mask is still not ready😭.

", "rawMarkdown": "I think we have to restart our work after the annotation mask is available. But the left time before deadline is less than 2 months, and the mask is still not ready😭.", "votes": 3 } ] }, { "id": 2517126, "postDate": "2023-11-08T08:33:29.973Z", "content": "

@masadia Thank you for your insights and tips.
\nThe approach of this competition is very different between with and without annotation masks. If you decide to add the annotation masks, we hope it can be provided as soon as possible. At least, please give a timeline. Because the datasets are very large, it takes a lot of time to train and tune the model. I think many participants are pending for the annotation masks.

", "rawMarkdown": "@masadia Thank you for your insights and tips. \nThe approach of this competition is very different between with and without annotation masks. If you decide to add the annotation masks, we hope it can be provided as soon as possible. At least, please give a timeline. Because the datasets are very large, it takes a lot of time to train and tune the model. I think many participants are pending for the annotation masks.", "votes": 13, "replies": [ { "id": 2524246, "postDate": "2023-11-14T05:11:37.173Z", "content": "

Any updates? @sohier @masadia

", "rawMarkdown": "Any updates? @sohier @masadia ", "votes": 10 } ] }, { "id": 2515382, "postDate": "2023-11-06T22:19:11.377Z", "content": "

@masadia
\nThanks Maryam for your video,
\nI'm still waiting for the pathologies annotations for 150 slides, and I would like to remind you to make them accessible :)

", "rawMarkdown": "@masadia \nThanks Maryam for your video,\nI'm still waiting for the pathologies annotations for 150 slides, and I would like to remind you to make them accessible :)", "votes": 6 }, { "id": 2518120, "postDate": "2023-11-09T04:22:16.040Z", "content": "

Can we please have a timeline and a confirmation of whether or when the annotation files will be shared? I stopped training my model since the announcement as the whole pipeline needs to be changed and it's been a week since the announcement and here have been no updates since.

", "rawMarkdown": "Can we please have a timeline and a confirmation of whether or when the annotation files will be shared? I stopped training my model since the announcement as the whole pipeline needs to be changed and it's been a week since the announcement and here have been no updates since. ", "votes": 5, "replies": [ { "id": 2518122, "postDate": "2023-11-09T04:26:09.807Z", "content": "

We're hoping to have an update to share either Friday afternoon PST or early next week.

", "rawMarkdown": "We're hoping to have an update to share either Friday afternoon PST or early next week.", "votes": 6, "replies": [ { "id": 2518776, "postDate": "2023-11-09T14:50:06.667Z", "content": "

Thank you @sohier for this info. Any plan to release TIFF/SVS images as well?

", "rawMarkdown": "Thank you @sohier for this info. Any plan to release TIFF/SVS images as well?", "votes": 5 }, { "id": 2529163, "postDate": "2023-11-18T02:10:37.980Z", "content": "

Hi @sohier , 9 days passed. Is there a deadline for the update of annotation?😭

", "rawMarkdown": "Hi @sohier , 9 days passed. Is there a deadline for the update of annotation?😭", "votes": 1 }, { "id": 2529352, "postDate": "2023-11-18T06:37:42.733Z", "content": "

the supplemental annotations have been provided, you can see in this pinned thread.

", "rawMarkdown": "the supplemental annotations have been provided, you can see in this [pinned thread](https://www.kaggle.com/competitions/UBC-OCEAN/discussion/455890).", "votes": 2 }, { "id": 2529433, "postDate": "2023-11-18T08:30:48.487Z", "content": "

Amazing! Got it!

", "rawMarkdown": "Amazing! Got it!" } ] } ] }, { "id": 2513778, "postDate": "2023-11-05T17:36:28.243Z", "content": "

@masadia is it be possible to share with us :

\n\n

Thanks a lot,
\nSimon

", "rawMarkdown": "@masadia is it be possible to share with us :\n\n- the proportion of outliers in the datasets\n- within the outliers, the proportions of rare cases vs normal cases vs non ovarian cases.\nThe strategies for the challenge won't be the same depending on your answer !\n\nThanks a lot,\nSimon", "votes": 3, "replies": [ { "id": 2516554, "postDate": "2023-11-07T19:08:21.537Z", "content": "

Hi Simon,

\n

Thanks for the question but hopefully you can appreciate that outlier detection is an important part of the problem at hand and that the goal would be a model which can generalize well, including situations where this information would not be known ahead of time.

\n

Best,
\nMichael

", "rawMarkdown": "Hi Simon,\n\nThanks for the question but hopefully you can appreciate that outlier detection is an important part of the problem at hand and that the goal would be a model which can generalize well, including situations where this information would not be known ahead of time.\n\nBest,\nMichael" } ] }, { "id": 2516624, "postDate": "2023-11-07T20:29:10.810Z", "content": "

Thank you Maryam for this excellent presentation.

", "rawMarkdown": "Thank you Maryam for this excellent presentation.", "votes": 1 }, { "id": 2515761, "postDate": "2023-11-07T08:10:37.930Z", "content": "

Hi @masadia , great presentation! Is there a overlap of medical center on training, public test and private test set? For example, if slide A in training set is produced by center1, is it possible a slide B produced in center1 occurred in public/private test set?

", "rawMarkdown": "Hi @masadia , great presentation! Is there a overlap of **medical center** on training, public test and private test set? For example, if slide A in training set is produced by center1, is it possible a slide B produced in center1 occurred in public/private test set?", "votes": 1 }, { "id": 2513691, "postDate": "2023-11-05T16:13:12.237Z", "content": "

thanks for the presentation. it's insightful and provides a good overview of the objective of the project.

", "rawMarkdown": "thanks for the presentation. it's insightful and provides a good overview of the objective of the project.", "votes": 1 }, { "id": 2517376, "postDate": "2023-11-08T12:36:09.697Z", "content": "

Does it means what I did in last month means nothing?

", "rawMarkdown": "Does it means what I did in last month means nothing?", "votes": 2 }, { "id": 2516815, "postDate": "2023-11-08T01:45:36.953Z", "content": "

If our team find some medical experts to assist us in manually annotating the provided training set, such as segmentation and detection tasks, is this allowed?

", "rawMarkdown": "If our team find some medical experts to assist us in manually annotating the provided training set, such as segmentation and detection tasks, is this allowed?", "votes": 2 }, { "id": 2510142, "postDate": "2023-11-02T18:44:18.200Z", "content": "

Thanks for the excellent video that puts the challenge into perspective and helps make it more meaningful in terms of what the technology can do to help people.

", "rawMarkdown": "Thanks for the excellent video that puts the challenge into perspective and helps make it more meaningful in terms of what the technology can do to help people.", "votes": 2 }, { "id": 2509720, "postDate": "2023-11-02T14:23:42.347Z", "content": "

Actually, quite informative presentation for a quick jump into a topic, segmentation data would be helpful

", "rawMarkdown": "Actually, quite informative presentation for a quick jump into a topic, segmentation data would be helpful ", "votes": 2 }, { "id": 3029692, "postDate": "2024-10-27T16:00:16.487Z", "content": "

There is an IEEE paper that is academic fraud. The title is \"OCEAN - Ovarian Cancer subtypE clAssification and outlier detectionioN using DenseNet121\". It claims that only convolutional networks were used without mentioning multi-instance learning. Then, 2,000 WSI images from the OECEN-UBC competition that our contestants could not get were used for training to obtain a classification accuracy of 99.7%. However, our first place winner only had a test accuracy of 0.6 and a 5-fold cross-validation accuracy of less than 90%\"\"\"\"

", "rawMarkdown": "There is an IEEE paper that is academic fraud. The title is \"OCEAN - Ovarian Cancer subtypE clAssification and outlier detectionioN using DenseNet121\". It claims that only convolutional networks were used without mentioning multi-instance learning. Then, 2,000 WSI images from the OECEN-UBC competition that our contestants could not get were used for training to obtain a classification accuracy of 99.7%. However, our first place winner only had a test accuracy of 0.6 and a 5-fold cross-validation accuracy of less than 90%![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16846649%2Fff519f283467f5bcf2e5416645a08cfa%2F_20241027234938.png?generation=1730044736614051&alt=media)![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16846649%2F2692818a6d51b91285ace54d23d8b383%2Ffake_acc.png?generation=1730044746592544&alt=media)" }, { "id": 2575184, "postDate": "2023-12-26T16:31:53.627Z", "content": "

Thank you Dear Maryam,
\nYour insights and tips have been incredibly helpful, and I'm excited to participate.

", "rawMarkdown": "Thank you Dear Maryam,\nYour insights and tips have been incredibly helpful, and I'm excited to participate." }, { "id": 2567547, "postDate": "2023-12-19T18:10:52.703Z", "content": "

Thank you for providing this insightful presentation, Maryam!

", "rawMarkdown": "Thank you for providing this insightful presentation, Maryam!" }, { "id": 2560413, "postDate": "2023-12-13T15:54:11.640Z", "content": "

Thank you for usefull information in a presentation, that helps to make a fast understanding of the topic to start NN training.

", "rawMarkdown": "Thank you for usefull information in a presentation, that helps to make a fast understanding of the topic to start NN training." }, { "id": 2554823, "postDate": "2023-12-09T12:42:04.917Z", "content": "

Thank you Maryam @masadia for the well-articulated presentation and the clarifications on the outlier detection requirement. 😊

", "rawMarkdown": "Thank you Maryam @masadia for the well-articulated presentation and the clarifications on the outlier detection requirement. 😊" }, { "id": 2550685, "postDate": "2023-12-06T08:07:06.737Z", "content": "

interesting

", "rawMarkdown": "interesting" }, { "id": 2527009, "postDate": "2023-11-16T08:01:32.053Z", "content": "

@masadia , Thank you for the presentation. It was very informative…😊
\nany updates on the annotation file's release?

", "rawMarkdown": "@masadia , Thank you for the presentation. It was very informative...😊\nany updates on the annotation file's release?" }, { "id": 2536608, "postDate": "2023-11-24T10:02:33.973Z", "content": "

Thank you for the presentation!

", "rawMarkdown": "Thank you for the presentation!" } ], "comments": [ { "id": 2508771, "author_name": "fam_taro", "author_url": "", "post_date": "2023-11-02T02:09:02.977000", "content": "

In general, it is not recommended to add a different format of data (tumor non-tumor mask in this case?) in the middle of a competition.

\n

This is because the data may require us to change our approach significantly, which can be very burdensome.

\n

I don't think this was the case in the PANDA competition.

", "votes": 17, "replies": [ { "id": 2513776, "author_name": "", "author_url": "", "post_date": "2023-11-05T17:36:08.810000", "content": "", "votes": 0, "replies": [] }, { "id": 2517052, "author_name": "ForcewithMe", "author_url": "", "post_date": "2023-11-08T07:16:19.320000", "content": "

I think we have to restart our work after the annotation mask is available. But the left time before deadline is less than 2 months, and the mask is still not ready😭.

", "votes": 3, "replies": [] } ] }, { "id": 2517126, "author_name": "Johnny Lee", "author_url": "", "post_date": "2023-11-08T08:33:29.973000", "content": "

@masadia Thank you for your insights and tips.
\nThe approach of this competition is very different between with and without annotation masks. If you decide to add the annotation masks, we hope it can be provided as soon as possible. At least, please give a timeline. Because the datasets are very large, it takes a lot of time to train and tune the model. I think many participants are pending for the annotation masks.

", "votes": 13, "replies": [ { "id": 2524246, "author_name": "Gunes Evitan", "author_url": "", "post_date": "2023-11-14T05:11:37.173000", "content": "

Any updates? @sohier @masadia

", "votes": 10, "replies": [] } ] }, { "id": 2515382, "author_name": "Somayyeh Gholami", "author_url": "", "post_date": "2023-11-06T22:19:11.377000", "content": "

@masadia
\nThanks Maryam for your video,
\nI'm still waiting for the pathologies annotations for 150 slides, and I would like to remind you to make them accessible :)

", "votes": 6, "replies": [] }, { "id": 2518120, "author_name": "Nima Ashjaee", "author_url": "", "post_date": "2023-11-09T04:22:16.040000", "content": "

Can we please have a timeline and a confirmation of whether or when the annotation files will be shared? I stopped training my model since the announcement as the whole pipeline needs to be changed and it's been a week since the announcement and here have been no updates since.

", "votes": 5, "replies": [ { "id": 2518122, "author_name": "Sohier Dane", "author_url": "", "post_date": "2023-11-09T04:26:09.807000", "content": "

We're hoping to have an update to share either Friday afternoon PST or early next week.

", "votes": 6, "replies": [ { "id": 2518776, "author_name": "jibounet", "author_url": "", "post_date": "2023-11-09T14:50:06.667000", "content": "

Thank you @sohier for this info. Any plan to release TIFF/SVS images as well?

", "votes": 5, "replies": [] }, { "id": 2529163, "author_name": "ForcewithMe", "author_url": "", "post_date": "2023-11-18T02:10:37.980000", "content": "

Hi @sohier , 9 days passed. Is there a deadline for the update of annotation?😭

", "votes": 1, "replies": [] }, { "id": 2529352, "author_name": "Sergey Kr", "author_url": "", "post_date": "2023-11-18T06:37:42.733000", "content": "

the supplemental annotations have been provided, you can see in this pinned thread.

", "votes": 2, "replies": [] }, { "id": 2529433, "author_name": "ForcewithMe", "author_url": "", "post_date": "2023-11-18T08:30:48.487000", "content": "

Amazing! Got it!

", "votes": 0, "replies": [] } ] } ] }, { "id": 2513778, "author_name": "simjeg", "author_url": "", "post_date": "2023-11-05T17:36:28.243000", "content": "

@masadia is it be possible to share with us :

\n\n

Thanks a lot,
\nSimon

", "votes": 3, "replies": [ { "id": 2516554, "author_name": "Michael Diaz-Stewart", "author_url": "", "post_date": "2023-11-07T19:08:21.537000", "content": "

Hi Simon,

\n

Thanks for the question but hopefully you can appreciate that outlier detection is an important part of the problem at hand and that the goal would be a model which can generalize well, including situations where this information would not be known ahead of time.

\n

Best,
\nMichael

", "votes": 0, "replies": [] } ] }, { "id": 2516624, "author_name": "Ons Loukil", "author_url": "", "post_date": "2023-11-07T20:29:10.810000", "content": "

Thank you Maryam for this excellent presentation.

", "votes": 1, "replies": [] }, { "id": 2515761, "author_name": "ForcewithMe", "author_url": "", "post_date": "2023-11-07T08:10:37.930000", "content": "

Hi @masadia , great presentation! Is there a overlap of medical center on training, public test and private test set? For example, if slide A in training set is produced by center1, is it possible a slide B produced in center1 occurred in public/private test set?

", "votes": 1, "replies": [] }, { "id": 2513691, "author_name": "Linkuz", "author_url": "", "post_date": "2023-11-05T16:13:12.237000", "content": "

thanks for the presentation. it's insightful and provides a good overview of the objective of the project.

", "votes": 1, "replies": [] }, { "id": 2517376, "author_name": "Sihan Lv", "author_url": "", "post_date": "2023-11-08T12:36:09.697000", "content": "

Does it means what I did in last month means nothing?

", "votes": 2, "replies": [] }, { "id": 2516815, "author_name": "Mr.Fire", "author_url": "", "post_date": "2023-11-08T01:45:36.953000", "content": "

If our team find some medical experts to assist us in manually annotating the provided training set, such as segmentation and detection tasks, is this allowed?

", "votes": 2, "replies": [] }, { "id": 2510142, "author_name": "Russ Tokuyama", "author_url": "", "post_date": "2023-11-02T18:44:18.200000", "content": "

Thanks for the excellent video that puts the challenge into perspective and helps make it more meaningful in terms of what the technology can do to help people.

", "votes": 2, "replies": [] }, { "id": 2509720, "author_name": "Aleksandr Lavrikov", "author_url": "", "post_date": "2023-11-02T14:23:42.347000", "content": "

Actually, quite informative presentation for a quick jump into a topic, segmentation data would be helpful

", "votes": 2, "replies": [] }, { "id": 3029692, "author_name": "Metavers", "author_url": "", "post_date": "2024-10-27T16:00:16.487000", "content": "

There is an IEEE paper that is academic fraud. The title is \"OCEAN - Ovarian Cancer subtypE clAssification and outlier detectionioN using DenseNet121\". It claims that only convolutional networks were used without mentioning multi-instance learning. Then, 2,000 WSI images from the OECEN-UBC competition that our contestants could not get were used for training to obtain a classification accuracy of 99.7%. However, our first place winner only had a test accuracy of 0.6 and a 5-fold cross-validation accuracy of less than 90%\"\"\"\"

", "votes": 0, "replies": [] }, { "id": 2575184, "author_name": "Sara Pouyan", "author_url": "", "post_date": "2023-12-26T16:31:53.627000", "content": "

Thank you Dear Maryam,
\nYour insights and tips have been incredibly helpful, and I'm excited to participate.

", "votes": 0, "replies": [] }, { "id": 2567547, "author_name": "T Leonidas", "author_url": "", "post_date": "2023-12-19T18:10:52.703000", "content": "

Thank you for providing this insightful presentation, Maryam!

", "votes": 0, "replies": [] }, { "id": 2560413, "author_name": "LovelyML", "author_url": "", "post_date": "2023-12-13T15:54:11.640000", "content": "

Thank you for usefull information in a presentation, that helps to make a fast understanding of the topic to start NN training.

", "votes": 0, "replies": [] }, { "id": 2554823, "author_name": "JK Ezhil", "author_url": "", "post_date": "2023-12-09T12:42:04.917000", "content": "

Thank you Maryam @masadia for the well-articulated presentation and the clarifications on the outlier detection requirement. 😊

", "votes": 0, "replies": [] }, { "id": 2550685, "author_name": "Stephan Schweitzer", "author_url": "", "post_date": "2023-12-06T08:07:06.737000", "content": "

interesting

", "votes": 0, "replies": [] }, { "id": 2527009, "author_name": "Nain_tiwari", "author_url": "", "post_date": "2023-11-16T08:01:32.053000", "content": "

@masadia , Thank you for the presentation. It was very informative…😊
\nany updates on the annotation file's release?

", "votes": 0, "replies": [] }, { "id": 2536608, "author_name": "Anastasia Gorinova", "author_url": "", "post_date": "2023-11-24T10:02:33.973000", "content": "

Thank you for the presentation!

", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "2508705": "Dear participants,\n\nWe're thrilled to share a presentation on the [UBC-OCEAN challenge](https://www.youtube.com/watch?v=49NP8VEOH5I), featuring essential dataset information and valuable tips. The annotation files mentioned in slides 12 and 13 (time: 8:45 to 10:15) have been submitted to Kaggle and will be accessible to you in a few days.\n\nWe remain dedicated to promptly providing the resources you need to enhance your model's performance.\n\nThank you for your participation, and please stay tuned for updates.\n\nBest regards,\nMaryam", "2508771": "In general, it is not recommended to add a different format of data (tumor non-tumor mask in this case?) in the middle of a competition. \n\nThis is because the data may require us to change our approach significantly, which can be very burdensome.\n\nI don't think this was the case in the PANDA competition.", "2517126": "@masadia Thank you for your insights and tips. \nThe approach of this competition is very different between with and without annotation masks. If you decide to add the annotation masks, we hope it can be provided as soon as possible. At least, please give a timeline. Because the datasets are very large, it takes a lot of time to train and tune the model. I think many participants are pending for the annotation masks.", "2515382": "@masadia \nThanks Maryam for your video,\nI'm still waiting for the pathologies annotations for 150 slides, and I would like to remind you to make them accessible :)", "2518120": "Can we please have a timeline and a confirmation of whether or when the annotation files will be shared? I stopped training my model since the announcement as the whole pipeline needs to be changed and it's been a week since the announcement and here have been no updates since. ", "2513778": "@masadia is it be possible to share with us :\n\n- the proportion of outliers in the datasets\n- within the outliers, the proportions of rare cases vs normal cases vs non ovarian cases.\nThe strategies for the challenge won't be the same depending on your answer !\n\nThanks a lot,\nSimon", "2516624": "Thank you Maryam for this excellent presentation.", "2515761": "Hi @masadia , great presentation! Is there a overlap of **medical center** on training, public test and private test set? For example, if slide A in training set is produced by center1, is it possible a slide B produced in center1 occurred in public/private test set?", "2513691": "thanks for the presentation. it's insightful and provides a good overview of the objective of the project.", "2517376": "Does it means what I did in last month means nothing?", "2516815": "If our team find some medical experts to assist us in manually annotating the provided training set, such as segmentation and detection tasks, is this allowed?", "2510142": "Thanks for the excellent video that puts the challenge into perspective and helps make it more meaningful in terms of what the technology can do to help people.", "2509720": "Actually, quite informative presentation for a quick jump into a topic, segmentation data would be helpful ", "3029692": "There is an IEEE paper that is academic fraud. The title is \"OCEAN - Ovarian Cancer subtypE clAssification and outlier detectionioN using DenseNet121\". It claims that only convolutional networks were used without mentioning multi-instance learning. Then, 2,000 WSI images from the OECEN-UBC competition that our contestants could not get were used for training to obtain a classification accuracy of 99.7%. However, our first place winner only had a test accuracy of 0.6 and a 5-fold cross-validation accuracy of less than 90%![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16846649%2Fff519f283467f5bcf2e5416645a08cfa%2F_20241027234938.png?generation=1730044736614051&alt=media)![](https://www.googleapis.com/download/storage/v1/b/kaggle-forum-message-attachments/o/inbox%2F16846649%2F2692818a6d51b91285ace54d23d8b383%2Ffake_acc.png?generation=1730044746592544&alt=media)", "2575184": "Thank you Dear Maryam,\nYour insights and tips have been incredibly helpful, and I'm excited to participate.", "2567547": "Thank you for providing this insightful presentation, Maryam!", "2560413": "Thank you for usefull information in a presentation, that helps to make a fast understanding of the topic to start NN training.", "2554823": "Thank you Maryam @masadia for the well-articulated presentation and the clarifications on the outlier detection requirement. 😊", "2550685": "interesting", "2527009": "@masadia , Thank you for the presentation. It was very informative...😊\nany updates on the annotation file's release?", "2536608": "Thank you for the presentation!" } }