Hi Guys!\nI really would have liked to achieve some nice result on this competition. Unfortunately I was motivated by some personal reason too.\nProbably I won't manage to achieve my original goal, but I'd like to make some tiny improvement in my work at least. But unfortunately either my current knowledge or some little latent technical issue (or both) held me up to do this so far.\nAt the start I worked with Hoyt Koepke's code and I had promising results and in addition I have a key idea I wanted to check, so I was hopeful enough. But later I have to change to some other platform mainly because of copyright problem, and I changed to nolearn. But for 2-3 weeks I'm just struggling with nolearn, most likely my bad, but I can't step forward a simple step. The top moment today was when I succeeded to reach to converge to MSE 0.25! :)) \nSo would anybody be so kind and help me to move away from this dead-lock with any kind of nolearn/lasagne convnet configuration(the least working is good for me at his monent) or/and some useful advise?\nI tried several data normalization, full army of net configuration (recall that with Hoyt's kind code I have good results) options but someway, somehow I am get stuck at this point. Mostly I had a learning curve, with decreasing training and very similarly increasing validation curve and I can't figure out why.\nThank you in advance, George
", "messages": [ { "id": "87016", "postDate": "07/26/2015 11:30:57", "content": "Hi Guys!\nI really would have liked to achieve some nice result on this competition. Unfortunately I was motivated by some personal reason too.\nProbably I won't manage to achieve my original goal, but I'd like to make some tiny improvement in my work at least. But unfortunately either my current knowledge or some little latent technical issue (or both) held me up to do this so far.\nAt the start I worked with Hoyt Koepke's code and I had promising results and in addition I have a key idea I wanted to check, so I was hopeful enough. But later I have to change to some other platform mainly because of copyright problem, and I changed to nolearn. But for 2-3 weeks I'm just struggling with nolearn, most likely my bad, but I can't step forward a simple step. The top moment today was when I succeeded to reach to converge to MSE 0.25! :)) \nSo would anybody be so kind and help me to move away from this dead-lock with any kind of nolearn/lasagne convnet configuration(the least working is good for me at his monent) or/and some useful advise?\nI tried several data normalization, full army of net configuration (recall that with Hoyt's kind code I have good results) options but someway, somehow I am get stuck at this point. Mostly I had a learning curve, with decreasing training and very similarly increasing validation curve and I can't figure out why.\nThank you in advance, George
", "rawMarkdown": "Hi Guys!\r\nI really would have liked to achieve some nice result on this competition. Unfortunately I was motivated by some personal reason too.\r\nProbably I won't manage to achieve my original goal, but I'd like to make some tiny improvement in my work at least. But unfortunately either my current knowledge or some little latent technical issue (or both) held me up to do this so far.\r\nAt the start I worked with Hoyt Koepke's code and I had promising results and in addition I have a key idea I wanted to check, so I was hopeful enough. But later I have to change to some other platform mainly because of copyright problem, and I changed to nolearn. But for 2-3 weeks I'm just struggling with nolearn, most likely my bad, but I can't step forward a simple step. The top moment today was when I succeeded to reach to converge to MSE 0.25! :)) \r\nSo would anybody be so kind and help me to move away from this dead-lock with any kind of nolearn/lasagne convnet configuration(the least working is good for me at his monent) or/and some useful advise?\r\nI tried several data normalization, full army of net configuration (recall that with Hoyt's kind code I have good results) options but someway, somehow I am get stuck at this point. Mostly I had a learning curve, with decreasing training and very similarly increasing validation curve and I can't figure out why.\r\nThank you in advance, George", "votes": null }, { "id": "87093", "postDate": "07/26/2015 22:24:54", "content": "Thank you guys...
", "rawMarkdown": "Thank you guys...", "votes": null }, { "id": "87099", "postDate": "07/26/2015 23:04:51", "content": "You with the downvote you're a gentleman... :))
", "rawMarkdown": "You with the downvote you're a gentleman... :))", "votes": null }, { "id": "87101", "postDate": "07/26/2015 23:19:39", "content": "Needed help and got downvotes that's a nice manner, but thank you again.
", "rawMarkdown": "Needed help and got downvotes that's a nice manner, but thank you again.", "votes": null }, { "id": "87140", "postDate": "07/27/2015 08:59:26", "content": "You with the downvotes, did you read it:
\n\n\n\n\n"Diabetic retinopathy is the leading cause of blindness in the\n working-age population of the developed world. It is estimated to\n affect over 93 million people."
\n
I don't really understand why this manner. Obviously - and you know well too - I haven't any chance to get with a help even in the top 200. But OK you don't help, no problem. You just are about pushing the button. \nAnyway I continuously work on the problem, I will solve for sure, I love challenges and will help this 93 million people (and their family...) and for the memory of my Mother who have died by the several effects of DR too. That was my last post here, thank you and good luck, have a nice day and have a nice - and hopefully fruitful - last day in this competion too!
", "rawMarkdown": "You with the downvotes, did you read it:\r\n\r\n> \"Diabetic retinopathy is the leading cause of blindness in the\r\n> working-age population of the developed world. It is estimated to\r\n> affect over 93 million people.\"\r\n\r\nI don't really understand why this manner. Obviously - and you know well too - I haven't any chance to get with a help even in the top 200. But OK you don't help, no problem. You just are about pushing the button. \r\nAnyway I continuously work on the problem, I will solve for sure, I love challenges and will help this **93 million people** (and their family...) and for the memory of **my Mother** who have died by the several effects of DR too. That was my last post here, thank you and good luck, have a nice day and have a nice - and hopefully fruitful - last day in this competion too!", "votes": null }, { "id": "87143", "postDate": "07/27/2015 09:21:38", "content": "Your question was probably too broad for the one guy who downvoted you.\nPeople probably didn't give you any hint for the same reason, your question is noy specific enough. Besides, there are not so many active contestants in this competition.
\n\nRegarding the downvotes to your other comments in your monologue , in my opinion, you probably deserved them. You didn't wait more than 10 hours before starting to get pushy and demanding.
\n\nBut don't get mad because of this. The downvoting has a purpose, but it can be ambigous many times.
\n\nAnd I am sure once the winning solution is published, you'll be much closer to your goal.\nCheers,
", "rawMarkdown": "Your question was probably too broad for the one guy who downvoted you.\r\nPeople probably didn't give you any hint for the same reason, your question is noy specific enough. Besides, there are not so many active contestants in this competition.\r\n\r\nRegarding the downvotes to your other comments in your monologue , in my opinion, you probably deserved them. You didn't wait more than 10 hours before starting to get pushy and demanding. \r\n\r\nBut don't get mad because of this. The downvoting has a purpose, but it can be ambigous many times.\r\n\r\nAnd I am sure once the winning solution is published, you'll be much closer to your goal.\r\nCheers,", "votes": null }, { "id": "87152", "postDate": "07/27/2015 11:44:47", "content": "Thanks Fernando your words. Note, that "You didn't wait more than 10 hours " is true, however I didn't because I couldn't because of shortness of the time we have to the end of the competition.. Furthermore, they had time to vote, hadn't they? But no time for help.\nAnyway thanks again for your time, I am over it and I look into the future for the achievements of DR related researches and for the other competitions too.\nHave a nice day! Cheers
", "rawMarkdown": "Thanks Fernando your words. Note, that \"You didn't wait more than 10 hours \" is true, however I didn't because I couldn't because of shortness of the time we have to the end of the competition.. Furthermore, they had time to vote, hadn't they? But no time for help.\r\nAnyway thanks again for your time, I am over it and I look into the future for the achievements of DR related researches and for the other competitions too.\r\nHave a nice day! Cheers", "votes": null } ], "comments": [ { "id": 87093, "author_name": "", "author_url": "", "post_date": "07/26/2015 22:24:54", "content": "Thank you guys...
", "votes": null, "replies": [] }, { "id": 87099, "author_name": "", "author_url": "", "post_date": "07/26/2015 23:04:51", "content": "You with the downvote you're a gentleman... :))
", "votes": null, "replies": [] }, { "id": 87101, "author_name": "", "author_url": "", "post_date": "07/26/2015 23:19:39", "content": "Needed help and got downvotes that's a nice manner, but thank you again.
", "votes": null, "replies": [] }, { "id": 87140, "author_name": "", "author_url": "", "post_date": "07/27/2015 08:59:26", "content": "You with the downvotes, did you read it:
\n\n\n\n\n"Diabetic retinopathy is the leading cause of blindness in the\n working-age population of the developed world. It is estimated to\n affect over 93 million people."
\n
I don't really understand why this manner. Obviously - and you know well too - I haven't any chance to get with a help even in the top 200. But OK you don't help, no problem. You just are about pushing the button. \nAnyway I continuously work on the problem, I will solve for sure, I love challenges and will help this 93 million people (and their family...) and for the memory of my Mother who have died by the several effects of DR too. That was my last post here, thank you and good luck, have a nice day and have a nice - and hopefully fruitful - last day in this competion too!
", "votes": null, "replies": [] }, { "id": 87143, "author_name": "fernandoconstantino", "author_url": "", "post_date": "07/27/2015 09:21:38", "content": "Your question was probably too broad for the one guy who downvoted you.\nPeople probably didn't give you any hint for the same reason, your question is noy specific enough. Besides, there are not so many active contestants in this competition.
\n\nRegarding the downvotes to your other comments in your monologue , in my opinion, you probably deserved them. You didn't wait more than 10 hours before starting to get pushy and demanding.
\n\nBut don't get mad because of this. The downvoting has a purpose, but it can be ambigous many times.
\n\nAnd I am sure once the winning solution is published, you'll be much closer to your goal.\nCheers,
", "votes": null, "replies": [] }, { "id": 87152, "author_name": "", "author_url": "", "post_date": "07/27/2015 11:44:47", "content": "Thanks Fernando your words. Note, that "You didn't wait more than 10 hours " is true, however I didn't because I couldn't because of shortness of the time we have to the end of the competition.. Furthermore, they had time to vote, hadn't they? But no time for help.\nAnyway thanks again for your time, I am over it and I look into the future for the achievements of DR related researches and for the other competitions too.\nHave a nice day! Cheers
", "votes": null, "replies": [] } ], "raw_markdown_by_id": { "87016": "Hi Guys!\r\nI really would have liked to achieve some nice result on this competition. Unfortunately I was motivated by some personal reason too.\r\nProbably I won't manage to achieve my original goal, but I'd like to make some tiny improvement in my work at least. But unfortunately either my current knowledge or some little latent technical issue (or both) held me up to do this so far.\r\nAt the start I worked with Hoyt Koepke's code and I had promising results and in addition I have a key idea I wanted to check, so I was hopeful enough. But later I have to change to some other platform mainly because of copyright problem, and I changed to nolearn. But for 2-3 weeks I'm just struggling with nolearn, most likely my bad, but I can't step forward a simple step. The top moment today was when I succeeded to reach to converge to MSE 0.25! :)) \r\nSo would anybody be so kind and help me to move away from this dead-lock with any kind of nolearn/lasagne convnet configuration(the least working is good for me at his monent) or/and some useful advise?\r\nI tried several data normalization, full army of net configuration (recall that with Hoyt's kind code I have good results) options but someway, somehow I am get stuck at this point. Mostly I had a learning curve, with decreasing training and very similarly increasing validation curve and I can't figure out why.\r\nThank you in advance, George", "87093": "Thank you guys...", "87099": "You with the downvote you're a gentleman... :))", "87101": "Needed help and got downvotes that's a nice manner, but thank you again.", "87140": "You with the downvotes, did you read it:\r\n\r\n> \"Diabetic retinopathy is the leading cause of blindness in the\r\n> working-age population of the developed world. It is estimated to\r\n> affect over 93 million people.\"\r\n\r\nI don't really understand why this manner. Obviously - and you know well too - I haven't any chance to get with a help even in the top 200. But OK you don't help, no problem. You just are about pushing the button. \r\nAnyway I continuously work on the problem, I will solve for sure, I love challenges and will help this **93 million people** (and their family...) and for the memory of **my Mother** who have died by the several effects of DR too. That was my last post here, thank you and good luck, have a nice day and have a nice - and hopefully fruitful - last day in this competion too!", "87143": "Your question was probably too broad for the one guy who downvoted you.\r\nPeople probably didn't give you any hint for the same reason, your question is noy specific enough. Besides, there are not so many active contestants in this competition.\r\n\r\nRegarding the downvotes to your other comments in your monologue , in my opinion, you probably deserved them. You didn't wait more than 10 hours before starting to get pushy and demanding. \r\n\r\nBut don't get mad because of this. The downvoting has a purpose, but it can be ambigous many times.\r\n\r\nAnd I am sure once the winning solution is published, you'll be much closer to your goal.\r\nCheers,", "87152": "Thanks Fernando your words. Note, that \"You didn't wait more than 10 hours \" is true, however I didn't because I couldn't because of shortness of the time we have to the end of the competition.. Furthermore, they had time to vote, hadn't they? But no time for help.\r\nAnyway thanks again for your time, I am over it and I look into the future for the achievements of DR related researches and for the other competitions too.\r\nHave a nice day! Cheers" }, "source": "meta" }