This just in. Conveniently, the first session is exactly relevant to this challenge, and those papers are listed first.
\n\n** ICASSP 2019 Open Preview Now Live **
\n\nConference papers are now live and freely accessible in IEEE Xplore® via Open Preview! \nhttps://signalprocessingsociety.us4.list-manage.com/track/click?u=25f275b5af555643e88abfbcb&id=f714ba2482&e=2b3d94157a\nThey are available publicly until the conference ends next Friday, 17 May, after which date they’ll be available only to IEEE Xplore® customers. Get a sneak peek at the cutting-edge signal processing research!
", "messages": [ { "id": "529818", "postDate": "05/10/2019 21:03:44", "content": "This just in. Conveniently, the first session is exactly relevant to this challenge, and those papers are listed first.
\n\n** ICASSP 2019 Open Preview Now Live **
\n\nConference papers are now live and freely accessible in IEEE Xplore® via Open Preview! \nhttps://signalprocessingsociety.us4.list-manage.com/track/click?u=25f275b5af555643e88abfbcb&id=f714ba2482&e=2b3d94157a\nThey are available publicly until the conference ends next Friday, 17 May, after which date they’ll be available only to IEEE Xplore® customers. Get a sneak peek at the cutting-edge signal processing research!
", "rawMarkdown": "This just in. Conveniently, the first session is exactly relevant to this challenge, and those papers are listed first.\n\n** ICASSP 2019 Open Preview Now Live **\n\nConference papers are now live and freely accessible in IEEE Xplore® via Open Preview! \nhttps://signalprocessingsociety.us4.list-manage.com/track/click?u=25f275b5af555643e88abfbcb&id=f714ba2482&e=2b3d94157a\nThey are available publicly until the conference ends next Friday, 17 May, after which date they’ll be available only to IEEE Xplore® customers. Get a sneak peek at the cutting-edge signal processing research!", "votes": null }, { "id": "530142", "postDate": "05/11/2019 21:56:41", "content": "Wow, today I've learned about prototypical networks :)\nHowever, based on that paper, VGG is still the best architecture for our case (I believe curated set contains ~75 samples per class). https://ieeexplore.ieee.org/document/8682591
", "rawMarkdown": "Wow, today I've learned about prototypical networks :)\nHowever, based on that paper, VGG is still the best architecture for our case (I believe curated set contains ~75 samples per class). https://ieeexplore.ieee.org/document/8682591", "votes": null }, { "id": "530177", "postDate": "05/12/2019 02:15:49", "content": "Regarding prototypical networks, I've tried in Humpback Whale Identification competition.\n- Github repo: https://github.com/daisukelab/protonet-fine-grained-clf\n- Discussion: https://www.kaggle.com/c/humpback-whale-identification/discussion/81085#latest-485567\nI haven't tried in this competition yet.
", "rawMarkdown": "Regarding prototypical networks, I've tried in [Humpback Whale Identification](https://www.kaggle.com/c/humpback-whale-identification) competition.\n- Github repo: https://github.com/daisukelab/protonet-fine-grained-clf\n- Discussion: https://www.kaggle.com/c/humpback-whale-identification/discussion/81085#latest-485567\nI haven't tried in this competition yet.", "votes": null }, { "id": "530257", "postDate": "05/12/2019 09:42:58", "content": "Also mentioned in that discussion is this paper about prototypical networks in a semi-supervised context that may also be interesting to try: http://metalearning.ml/2017/papers/metalearn17_boney.pdf
", "rawMarkdown": "Also mentioned in that discussion is this paper about prototypical networks in a semi-supervised context that may also be interesting to try: http://metalearning.ml/2017/papers/metalearn17_boney.pdf", "votes": null } ], "comments": [ { "id": 530142, "author_name": "vandalko", "author_url": "", "post_date": "05/11/2019 21:56:41", "content": "Wow, today I've learned about prototypical networks :)\nHowever, based on that paper, VGG is still the best architecture for our case (I believe curated set contains ~75 samples per class). https://ieeexplore.ieee.org/document/8682591
", "votes": null, "replies": [ { "id": 530177, "author_name": "daisukelab", "author_url": "", "post_date": "05/12/2019 02:15:49", "content": "Regarding prototypical networks, I've tried in Humpback Whale Identification competition.\n- Github repo: https://github.com/daisukelab/protonet-fine-grained-clf\n- Discussion: https://www.kaggle.com/c/humpback-whale-identification/discussion/81085#latest-485567\nI haven't tried in this competition yet.
", "votes": null, "replies": [] }, { "id": 530257, "author_name": "mnpinto", "author_url": "", "post_date": "05/12/2019 09:42:58", "content": "Also mentioned in that discussion is this paper about prototypical networks in a semi-supervised context that may also be interesting to try: http://metalearning.ml/2017/papers/metalearn17_boney.pdf
", "votes": null, "replies": [] } ] } ], "raw_markdown_by_id": { "529818": "This just in. Conveniently, the first session is exactly relevant to this challenge, and those papers are listed first.\n\n** ICASSP 2019 Open Preview Now Live **\n\nConference papers are now live and freely accessible in IEEE Xplore® via Open Preview! \nhttps://signalprocessingsociety.us4.list-manage.com/track/click?u=25f275b5af555643e88abfbcb&id=f714ba2482&e=2b3d94157a\nThey are available publicly until the conference ends next Friday, 17 May, after which date they’ll be available only to IEEE Xplore® customers. Get a sneak peek at the cutting-edge signal processing research!", "530142": "Wow, today I've learned about prototypical networks :)\nHowever, based on that paper, VGG is still the best architecture for our case (I believe curated set contains ~75 samples per class). https://ieeexplore.ieee.org/document/8682591", "530177": "Regarding prototypical networks, I've tried in [Humpback Whale Identification](https://www.kaggle.com/c/humpback-whale-identification) competition.\n- Github repo: https://github.com/daisukelab/protonet-fine-grained-clf\n- Discussion: https://www.kaggle.com/c/humpback-whale-identification/discussion/81085#latest-485567\nI haven't tried in this competition yet.", "530257": "Also mentioned in that discussion is this paper about prototypical networks in a semi-supervised context that may also be interesting to try: http://metalearning.ml/2017/papers/metalearn17_boney.pdf" }, "source": "meta" }