![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2779868%2F94c3c03cdf80335e400195c1e019eef4%2FAnnotation%202020-03-23%20193632.png?generation=1584992242431582&alt=media\")
The question would be: Why no points or prizes? \nNo points + quota constraints ... I can't understand.\nI don't understand this change of policy from last year.
\n\nLast year iMet, iMaterialist, Kuzushiji Recognition, Recursion Cellular, Human Atlas, all the Google comps! I think those research competitions were great, a lot of people competed and for that reason the solutions were great too, as the rewards :)
\n\nThe competition: iMaterialist Challenge on Product Recognition at FGVC6, CVPR2019 had a prize pool of 3000$ Many competitions from FGCV6 awarded money and/or points.
\n\nLast year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO for 1 reason: the best kagglers were competing and the result was amazing (as you can check at the LB)
\n\n
Unless you have plenty of resources or this is your research field, I don't see how these competitions can encourage/motivate competitors and CV experts to solve these problems...
\n\nI was competing here last year with the purpose of learning new segmentation problems. Kuzushiji Recognition is another research competition that encouraged participants not with points, but great prizes.\nI think Kaggle is not only a place for learning/playing with data and solve small problems as many people think. Here I have read complete, ingenious and novel solutions for real-world complex problems, always close to the SOTA. If you want such solutions, you'd better to offer something too.
\n\nPeople is free to join or not depending on how well organized is the comp, the timing, the topic and imo one of the most important factors are the prizes. An easy way to encourage/motivate top kagglers who can provide such solutions (and will do in many cases) is to offer great prizes (for example Kuzushiji).
\n\nThese competitions shouldn't be a \"Knowledge\" competition, probably an \"InClass\" competition or just upload the dataset with tasks.
\n\nI think the main point, the reason why kaggle works is because a lot of people compete that's the point imo. Motivate people to join, get interested in the subject and compete… and obviously have fun!
\n\nJust saying. As always I'd love to read your opinion.
", "messages": [ { "id": 783935, "postDate": "2020-03-23T20:01:23.437Z", "content": "The question would be: Why no points or prizes? \nNo points + quota constraints ... I can't understand.\nI don't understand this change of policy from last year.
\n\nLast year iMet, iMaterialist, Kuzushiji Recognition, Recursion Cellular, Human Atlas, all the Google comps! I think those research competitions were great, a lot of people competed and for that reason the solutions were great too, as the rewards :)
\n\nThe competition: iMaterialist Challenge on Product Recognition at FGVC6, CVPR2019 had a prize pool of 3000$ Many competitions from FGCV6 awarded money and/or points.
\n\nLast year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO for 1 reason: the best kagglers were competing and the result was amazing (as you can check at the LB)
\n\n
Unless you have plenty of resources or this is your research field, I don't see how these competitions can encourage/motivate competitors and CV experts to solve these problems...
\n\nI was competing here last year with the purpose of learning new segmentation problems. Kuzushiji Recognition is another research competition that encouraged participants not with points, but great prizes.\nI think Kaggle is not only a place for learning/playing with data and solve small problems as many people think. Here I have read complete, ingenious and novel solutions for real-world complex problems, always close to the SOTA. If you want such solutions, you'd better to offer something too.
\n\nPeople is free to join or not depending on how well organized is the comp, the timing, the topic and imo one of the most important factors are the prizes. An easy way to encourage/motivate top kagglers who can provide such solutions (and will do in many cases) is to offer great prizes (for example Kuzushiji).
\n\nThese competitions shouldn't be a \"Knowledge\" competition, probably an \"InClass\" competition or just upload the dataset with tasks.
\n\nI think the main point, the reason why kaggle works is because a lot of people compete that's the point imo. Motivate people to join, get interested in the subject and compete… and obviously have fun!
\n\nJust saying. As always I'd love to read your opinion.
", "rawMarkdown": "The question would be: **Why no points or prizes?** \nNo points + quota constraints ... I can't understand.\nI don't understand this change of policy from last year. \n\nLast year iMet, iMaterialist, Kuzushiji Recognition, Recursion Cellular, Human Atlas, all the Google comps! I think those research competitions were great, a lot of people **competed** and for that reason the solutions were great too, **as the rewards** :)\n\nThe competition: iMaterialist Challenge on Product Recognition at FGVC6, CVPR2019 had a prize pool of 3000$ Many competitions from FGCV6 awarded money and/or points.\n\nLast year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO for 1 reason: the best kagglers were competing and the result was amazing (as you can check at the LB)\n\n\n\nUnless you have plenty of resources or this is your research field, I don't see how these competitions can encourage/motivate competitors and CV experts to solve these problems...\n\nI was competing here last year with the purpose of learning new segmentation problems. [Kuzushiji Recognition](https://www.kaggle.com/c/kuzushiji-recognition/) is another research competition that encouraged participants not with points, but great prizes.\nI think Kaggle is **not only** a place for learning/playing with data and solve small problems as many people think. Here I have read complete, ingenious and novel solutions for real-world complex problems, always close to the SOTA. If you want such solutions, you'd better to offer something too. \n\nPeople is free to join or not depending on how well organized is the comp, the timing, the topic and imo one of the most important factors are the prizes. An easy way to encourage/motivate top kagglers who can provide such solutions (and will do in many cases) is to offer great prizes (for example Kuzushiji).\n\nThese competitions shouldn't be a \"Knowledge\" competition, probably an \"InClass\" competition or just upload the dataset with tasks.\n\nI think the main point, the reason why kaggle works is because **a lot of people compete** that's the point imo. Motivate people to join, get interested in the subject and compete… and obviously have fun!\n\nJust saying. As always I'd love to read your opinion.", "votes": 61 }, { "id": 786025, "postDate": "2020-03-25T15:27:42.187Z", "content": "Maybe the prize could be at least accommodation and registration cost for the conference?
", "rawMarkdown": "Maybe the prize could be at least accommodation and registration cost for the conference? ", "votes": 6, "replies": [ { "id": 823151, "postDate": "2020-04-27T13:00:15.580Z", "content": "Yes - granted, I'm not in academia, but is the possibility that you may get to present at a conference, for which you'd have to self-fund accommodation/travel, really that attractive?
", "rawMarkdown": "Yes - granted, I'm not in academia, but is the possibility that you _may_ get to present at a conference, for which you'd have to self-fund accommodation/travel, really that attractive?" } ] }, { "id": 784901, "postDate": "2020-03-24T15:37:39.853Z", "content": ">Last year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO
\n\nI don't agree with you. In the last year, we had a very bad experience with X5 people in Ifashion and Imet competition.
", "rawMarkdown": ">Last year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO\n\nI don't agree with you. In the last year, we had a very bad experience with X5 people in Ifashion and Imet competition.", "votes": 5, "replies": [ { "id": 785007, "postDate": "2020-03-24T17:41:24.767Z", "content": "yes you are right, but there were many other research competitions and great results. And even after that, I remember great solutions from many kagglers non related with X5. For example iMet 2019 winner. Moreover, the problem at iMet 2019 is really famous, but a lot of people competed! that's the point imo. Motivate people to join, get interested in the subject and compete... and obviously have fun!
\n\n![\"\"](\"https://www.googleapis.com/download/storage/v1/b/kaggle-user-content/o/inbox%2F2779868%2Fc84ffcdd81ae989e6d9aecad8b43e6ac%2FAnnotation%202020-03-24%20173759.png?generation=1585071515173688&alt=media\")
@rinnqd @jesucristo what happened last year, just out of interest?
", "rawMarkdown": "@rinnqd @jesucristo what happened last year, just out of interest?" } ] }, { "id": 784263, "postDate": "2020-03-24T03:54:45.157Z", "content": "What do organizers expect from a research competition?. \nIMO, they are:
\n\nLook at the previous year.
\n\nThe top solutions used current SOTA Network to win. Of course, you will be rejected immediately if you submit the paper :)). \"What is news\", \"No novelty\", \"What is your motivation?\" are the sentences I often receive from the review comments :D.
Most solutions are ensemble-based. In academia, we prefer single models than ensemble models.
In Grand-challenge, codalab or other platforms, I believe that there are very rare competitions having prizes money. But some of them has much more participants than you think (up to few hundred). They work for research purposes and for their passion. The researchers consider papers more than money.
\n\nSo please dont claim that No points/prizes are killing research competitions :D.
\nInstead, I think Kagglers are ruining the meaning of research competition
Hi @backaggle , I hope you are doing great! I agree with you in most of the points. But I don't understand this change of policy from last year. I think Kagglers are ruining the meaning of research competition people is free to join or not depending on how well organized is the comp, the timing, the topic and imo one of the most important factors are the prizes. An easy way to encourage/motivate top kagglers who can provide such solutions (and will do in many cases) is to offer great prizes (for example Kuzushiji).
If the idea is you have 2 months, please provide novel ideas/architectures and please spend a couple of weeks (at least) preparing a paper it is OK, but why last year the policy was different (and right in my opinion)? why this change?. I don't think they're opposing ideas, last year iMet, iMaterialist, Kuzushiji Recognition, Recursion Cellular, Human Atlas, all the Google comps! I think those research competitions were great, a lot of people competed and for that reason the solutions were great too, as the rewards :)
Hi @jesucristo ,
\n\n\n\n\none of the most important factors are the prizes.
\n
I agree that this point will encourage more people to join.
\n\n\n\n\nAn easy way to encourage/motivate top kagglers
\n
Yes, it will attract many top kagglers to provide the great solution You might think that it is useful for the organizers but actually not. It is meaningless if it is not presented to the science sociality. I met some friends at ICCV last year and told about Kaggle. Surprising to me that they dont know what is Kaggle at all even they wrote great papers in there. Even they know about this great solution, do you think citing a writeup solution is normal?
", "rawMarkdown": "Hi @jesucristo , \n> one of the most important factors are the prizes. \n\nI agree that this point will encourage more people to join. \n\n> An easy way to encourage/motivate top kagglers \n\n Yes, it will attract many top kagglers to provide the **great** solution You might think that it is useful for the organizers but actually not. It is meaningless if it is not presented to the science sociality. I met some friends at ICCV last year and told about Kaggle. Surprising to me that they dont know what is Kaggle at all even they wrote great papers in there. Even they know about this great solution, do you think citing a writeup solution is normal? ", "votes": 2 }, { "id": 784874, "postDate": "2020-03-24T14:55:45.503Z", "rawMarkdown": "", "isDeleted": true }, { "id": 785016, "postDate": "2020-03-24T17:49:35.213Z", "content": "It is meaningless if it is not presented to the science sociality I completely agree with you :) and that's why I'd like to see more research comps. At google anual comps or Kuzushiji, the prizes main purpose was to cover expenses (I guess). Let's see what they decide for the iMet 2020 competition.
@backaggle @rinnqd Looks like this is the new policy, iMet 2020 has no prizes or points... sad, but I respect it.
", "rawMarkdown": "@backaggle @rinnqd Looks like this is the new policy, iMet 2020 has no prizes or points... sad, but I respect it." }, { "id": 788744, "postDate": "2020-03-28T00:40:26.267Z", "content": "@jesucristo you're right this is so sad! They can at least encourage young talents with the accommodation, flight and the registration cost as @salmanhiro said
", "rawMarkdown": "@jesucristo you're right this is so sad! They can at least encourage young talents with the accommodation, flight and the registration cost as @salmanhiro said ", "votes": 2 }, { "id": 789974, "postDate": "2020-03-29T07:14:27.953Z", "content": "@jesucristo @rinnqd The problem is, as I think, last year's fiasco. They can atleast provide medals/registration cost but they do not need to provide cash.
", "rawMarkdown": "@jesucristo @rinnqd The problem is, as I think, last year's fiasco. They can atleast provide medals/registration cost but they do not need to provide cash.", "votes": 2 }, { "id": 830324, "postDate": "2020-05-02T14:06:04.360Z", "content": "@nxrprime what was last year's fiasco?
", "rawMarkdown": "@nxrprime what was last year's fiasco?" }, { "id": 830330, "postDate": "2020-05-02T14:11:43.480Z", "content": "@marcogorelli there were evidences that some people scrapped the data (real labels) and used them to boost the score (against the rules)
", "rawMarkdown": "@marcogorelli there were evidences that some people scrapped the data (real labels) and used them to boost the score (against the rules)", "votes": 1 } ] }, { "id": 793642, "postDate": "2020-04-01T07:13:57.170Z", "content": "no prizes or points, why?
", "rawMarkdown": "no prizes or points, why?" } ], "comments": [ { "id": 786025, "author_name": "Salman Chen", "author_url": "", "post_date": "2020-03-25T15:27:42.187000", "content": "Maybe the prize could be at least accommodation and registration cost for the conference?
", "votes": 6, "replies": [ { "id": 823151, "author_name": "Marco Gorelli", "author_url": "", "post_date": "2020-04-27T13:00:15.580000", "content": "Yes - granted, I'm not in academia, but is the possibility that you may get to present at a conference, for which you'd have to self-fund accommodation/travel, really that attractive?
", "votes": 0, "replies": [] } ] }, { "id": 784901, "author_name": "Firas Baba", "author_url": "", "post_date": "2020-03-24T15:37:39.853000", "content": ">Last year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO
\n\nI don't agree with you. In the last year, we had a very bad experience with X5 people in Ifashion and Imet competition.
", "votes": 5, "replies": [ { "id": 785007, "author_name": "Nanashi", "author_url": "", "post_date": "2020-03-24T17:41:24.767000", "content": "yes you are right, but there were many other research competitions and great results. And even after that, I remember great solutions from many kagglers non related with X5. For example iMet 2019 winner. Moreover, the problem at iMet 2019 is really famous, but a lot of people competed! that's the point imo. Motivate people to join, get interested in the subject and compete... and obviously have fun!
\n\n
@rinnqd @jesucristo what happened last year, just out of interest?
", "votes": 0, "replies": [] } ] }, { "id": 784263, "author_name": "cab", "author_url": "", "post_date": "2020-03-24T03:54:45.157000", "content": "What do organizers expect from a research competition?. \nIMO, they are:
\n\nLook at the previous year.
\n\nThe top solutions used current SOTA Network to win. Of course, you will be rejected immediately if you submit the paper :)). \"What is news\", \"No novelty\", \"What is your motivation?\" are the sentences I often receive from the review comments :D.
Most solutions are ensemble-based. In academia, we prefer single models than ensemble models.
In Grand-challenge, codalab or other platforms, I believe that there are very rare competitions having prizes money. But some of them has much more participants than you think (up to few hundred). They work for research purposes and for their passion. The researchers consider papers more than money.
\n\nSo please dont claim that No points/prizes are killing research competitions :D.
\nInstead, I think Kagglers are ruining the meaning of research competition
Hi @backaggle , I hope you are doing great! I agree with you in most of the points. But I don't understand this change of policy from last year. I think Kagglers are ruining the meaning of research competition people is free to join or not depending on how well organized is the comp, the timing, the topic and imo one of the most important factors are the prizes. An easy way to encourage/motivate top kagglers who can provide such solutions (and will do in many cases) is to offer great prizes (for example Kuzushiji).
If the idea is you have 2 months, please provide novel ideas/architectures and please spend a couple of weeks (at least) preparing a paper it is OK, but why last year the policy was different (and right in my opinion)? why this change?. I don't think they're opposing ideas, last year iMet, iMaterialist, Kuzushiji Recognition, Recursion Cellular, Human Atlas, all the Google comps! I think those research competitions were great, a lot of people competed and for that reason the solutions were great too, as the rewards :)
Hi @jesucristo ,
\n\n\n\n\none of the most important factors are the prizes.
\n
I agree that this point will encourage more people to join.
\n\n\n\n\nAn easy way to encourage/motivate top kagglers
\n
Yes, it will attract many top kagglers to provide the great solution You might think that it is useful for the organizers but actually not. It is meaningless if it is not presented to the science sociality. I met some friends at ICCV last year and told about Kaggle. Surprising to me that they dont know what is Kaggle at all even they wrote great papers in there. Even they know about this great solution, do you think citing a writeup solution is normal?
", "votes": 2, "replies": [] }, { "id": 784874, "author_name": "", "author_url": "", "post_date": "2020-03-24T14:55:45.503000", "content": "", "votes": 0, "replies": [] }, { "id": 785016, "author_name": "Nanashi", "author_url": "", "post_date": "2020-03-24T17:49:35.213000", "content": "It is meaningless if it is not presented to the science sociality I completely agree with you :) and that's why I'd like to see more research comps. At google anual comps or Kuzushiji, the prizes main purpose was to cover expenses (I guess). Let's see what they decide for the iMet 2020 competition.
@backaggle @rinnqd Looks like this is the new policy, iMet 2020 has no prizes or points... sad, but I respect it.
", "votes": 0, "replies": [] }, { "id": 788744, "author_name": "Firas Baba", "author_url": "", "post_date": "2020-03-28T00:40:26.267000", "content": "@jesucristo you're right this is so sad! They can at least encourage young talents with the accommodation, flight and the registration cost as @salmanhiro said
", "votes": 2, "replies": [] }, { "id": 789974, "author_name": "Trigram", "author_url": "", "post_date": "2020-03-29T07:14:27.953000", "content": "@jesucristo @rinnqd The problem is, as I think, last year's fiasco. They can atleast provide medals/registration cost but they do not need to provide cash.
", "votes": 2, "replies": [] }, { "id": 830324, "author_name": "Marco Gorelli", "author_url": "", "post_date": "2020-05-02T14:06:04.360000", "content": "@nxrprime what was last year's fiasco?
", "votes": 0, "replies": [] }, { "id": 830330, "author_name": "Nanashi", "author_url": "", "post_date": "2020-05-02T14:11:43.480000", "content": "@marcogorelli there were evidences that some people scrapped the data (real labels) and used them to boost the score (against the rules)
", "votes": 1, "replies": [] } ] }, { "id": 793642, "author_name": "Harland_Lau", "author_url": "", "post_date": "2020-04-01T07:13:57.170000", "content": "no prizes or points, why?
", "votes": 0, "replies": [] } ], "raw_markdown_by_id": { "783935": "The question would be: **Why no points or prizes?** \nNo points + quota constraints ... I can't understand.\nI don't understand this change of policy from last year. \n\nLast year iMet, iMaterialist, Kuzushiji Recognition, Recursion Cellular, Human Atlas, all the Google comps! I think those research competitions were great, a lot of people **competed** and for that reason the solutions were great too, **as the rewards** :)\n\nThe competition: iMaterialist Challenge on Product Recognition at FGVC6, CVPR2019 had a prize pool of 3000$ Many competitions from FGCV6 awarded money and/or points.\n\nLast year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO for 1 reason: the best kagglers were competing and the result was amazing (as you can check at the LB)\n\n\n\nUnless you have plenty of resources or this is your research field, I don't see how these competitions can encourage/motivate competitors and CV experts to solve these problems...\n\nI was competing here last year with the purpose of learning new segmentation problems. [Kuzushiji Recognition](https://www.kaggle.com/c/kuzushiji-recognition/) is another research competition that encouraged participants not with points, but great prizes.\nI think Kaggle is **not only** a place for learning/playing with data and solve small problems as many people think. Here I have read complete, ingenious and novel solutions for real-world complex problems, always close to the SOTA. If you want such solutions, you'd better to offer something too. \n\nPeople is free to join or not depending on how well organized is the comp, the timing, the topic and imo one of the most important factors are the prizes. An easy way to encourage/motivate top kagglers who can provide such solutions (and will do in many cases) is to offer great prizes (for example Kuzushiji).\n\nThese competitions shouldn't be a \"Knowledge\" competition, probably an \"InClass\" competition or just upload the dataset with tasks.\n\nI think the main point, the reason why kaggle works is because **a lot of people compete** that's the point imo. Motivate people to join, get interested in the subject and compete… and obviously have fun!\n\nJust saying. As always I'd love to read your opinion.", "786025": "Maybe the prize could be at least accommodation and registration cost for the conference? ", "784901": ">Last year this competition (iMaterialist (Fashion) 2019 at FGVC6) was great IMHO\n\nI don't agree with you. In the last year, we had a very bad experience with X5 people in Ifashion and Imet competition.", "784263": "\nWhat do organizers expect from a research competition?. \nIMO, they are: \n\n- Novel ideas/architectures \n- Someone is really willing to write/submit the paper and join the conference/workshop\n\nLook at the previous year. \n\n- The top solutions used current SOTA Network to win. Of course, you will be rejected immediately if you submit the paper :)). \"What is news\", \"No novelty\", \"What is your motivation?\" are the sentences I often receive from the review comments :D. \n\n- Most solutions are ensemble-based. In academia, we prefer single models than ensemble models. \n- Not so many participants wanted to join the conference/workshop. I remember that the organizers had to ask every single person to join the conference or not. \n\nIn Grand-challenge, codalab or other platforms, I believe that there are very rare competitions having prizes money. But some of them has much more participants than you think (up to few hundred). They work for research purposes and for their passion. The researchers consider papers more than money. \n\nSo please dont claim that **No points/prizes are killing research competitions** :D. \nInstead, I think Kagglers are ruining the meaning of research competition", "793642": "no prizes or points, why?" } }