logo资料库

自监督学习教程（来源于牛津大学&DeepMind）.pdf

发布时间：2022-06-19 发布人：admin 分类：说明书 资料大小：8.97M 资料格式：pdf 举报 版权申诉
第1页 / 共141页
第2页 / 共141页
第3页 / 共141页
第4页 / 共141页
第5页 / 共141页
第6页 / 共141页
第7页 / 共141页
第8页 / 共141页
资料共141页，剩余部分请下载后查看
    Self-Supervised Learning Andrew Zisserman July 2019 Slides from: Carl Doersch, Ishan Misra, Andrew Owens, AJ Piergiovanni, Carl Vondrick, Richard Zhang
    The ImageNet Challenge Story … 1000 categories • Training: 1000 images for each category • Testing: 100k images
    The ImageNet Challenge Story … strong supervision
    The ImageNet Challenge Story … outcomes Strong supervision: • Features from networks trained on ImageNet can be used for other visual tasks, e.g. detection, segmentation, action recognition, fine grained visual classification • To some extent, any visual task can be solved now by: 1. Construct a large-scale dataset labelled for that task 2. Specify a training loss and neural network architecture 3. Train the network and deploy • Are there alternatives to strong supervision for training? Self-Supervised learning ….
    Why Self-Supervision? 1. Expense of producing a new dataset for each new task 2. Some areas are supervision-starved, e.g. medical data, where it is hard to obtain annotation 3. Untapped/availability of vast numbers of unlabelled images/videos – Facebook: one billion images uploaded per day – 300 hours of video are uploaded to YouTube every minute 4. How infants may learn …
    Self-Supervised Learning The Scientist in the Crib: What Early Learning Tells Us About the Mind by Alison Gopnik, Andrew N. Meltzoff and Patricia K. Kuhl The Development of Embodied Cognition: Six Lessons from Babies by Linda Smith and Michael Gasser
    What is Self-Supervision? • A form of unsupervised learning where the data provides the supervision • In general, withhold some part of the data, and task the network with predicting it • The task defines a proxy loss, and the network is forced to learn what we really care about, e.g. a semantic representation, in order to solve it • In recent work we might also choose tasks that we care about ….
    Example: relative positioning Train network to predict relative position of two regions in the same image 8 possible locations Classifier CNN CNN Randomly Sample Patch Sample Second Patch Unsupervised visual representation learning by context prediction, Carl Doersch, Abhinav Gupta, Alexei A. Efros, ICCV 2015
    分享到：
    赞 收藏

    相关推荐

    资料库

    人工智能

    共收录2047份资料，累计62个分类，关注成员有19位，主要包括：GNSS,Pico-SIM868-GSM,GPRS,NB-IoT,GPRSHAT,SIM7070GCat-M,GNSSHAT,GSM,NB-IoTHAT,SIM7080GCat-M,R800CGSM,SIM800CGSM,UPSPowerModule(B),UPSPowerModule,Pico-SIM7020C-NB-IoT,SIM7028NB-IoTHAT,SIM7020CNB-IoTHAT,SIM7000XNB-IoTHAT,A7670CCat-1HAT,A7600C1Cat-1,SIM7600X4G&LTECat-1HAT,SIM7600X4GDONGLE用户手册,RM520N-GL5GHAT,RM500Q-GL5GHAT,RM500U-CN5GHAT,SIM820X-M25GforJetsonNano,AudioCardforJetsonNano,SIM7600G-H4GforJetsonNano,OpenNCC-Lite-4G8M,IMX219-77Camera,13.3inche-PaperHAT,10.3inche-PaperHAT(D),10.3inche-PaperHAT,9.7inche-PaperHAT,7.8inche-PaperHAT,6inchHDTouche-PaperHAT,6inchHDe-PaperHAT,6inche-PaperHAT,5.83inche-PaperHAT(C),LCD-Clock-A,0.95inchRGBOLED(A),0.96inchRGBOLEDModule,0.96inchOLEDModule,0.96inchOLED(A),1.27inchRGBOLEDModule,1.3inchOLEDModule(C),1.3inchOLED(A),1.32inchOLEDModule,1.5inchRGBOLEDModule,1.5inchOLEDModule,JETSONORINNX,自然语言处理,搜索引擎,VR,计算广告,深度学习,机器学习

    热门标签

    GNSS Pico-SIM868-GSM GPRS NB-IoT GPRSHAT SIM7070GCat-M GNSSHAT GSM NB-IoTHAT SIM7080GCat-M R800CGSM SIM800CGSM UPSPowerModule(B) UPSPowerModule Pico-SIM7020C-NB-IoT SIM7028NB-IoTHAT SIM7020CNB-IoTHAT SIM7000XNB-IoTHAT A7670CCat-1HAT A7600C1Cat-1 SIM7600X4G&LTECat-1HAT SIM7600X4GDONGLE用户手册 RM520N-GL5GHAT RM500Q-GL5GHAT RM500U-CN5GHAT SIM820X-M25GforJetsonNano AudioCardforJetsonNano SIM7600G-H4GforJetsonNano OpenNCC-Lite-4G8M IMX219-77Camera 13.3inche-PaperHAT 10.3inche-PaperHAT(D) 10.3inche-PaperHAT 9.7inche-PaperHAT 7.8inche-PaperHAT 6inchHDTouche-PaperHAT 6inchHDe-PaperHAT 6inche-PaperHAT 5.83inche-PaperHAT(C) LCD-Clock-A 0.95inchRGBOLED(A) 0.96inchRGBOLEDModule 0.96inchOLEDModule 0.96inchOLED(A) 1.27inchRGBOLEDModule 1.3inchOLEDModule(C) 1.3inchOLED(A) 1.32inchOLEDModule 1.5inchRGBOLEDModule 1.5inchOLEDModule JETSONORINNX 自然语言处理 搜索引擎 VR 计算广告 深度学习 机器学习

    最新资料