Guoxian Dai

Unknown words tsunami : 海啸 hyperbolic : 曲线的；夸张的 supersede : 取代；超越 meteoric : 大气的；流星的；极速的 reign : 统治；支配 unyielding : 不服的；坚强的 redeem : 救赎；挽回 primer : 初级读本；原始物 wimpy : 懦弱的；无用的 nemesis : 复仇女神；报应 folly : 愚蠢；荒唐 ostensibly : 表面；外表 synapse : 突触 irk : 惹恼；厌烦 SNARC hype : 大肆宣传；皮下兴奋 thaw : 溶解；变暖和 oval : 椭圆 folly : 愚蠢；荒唐 ostensibly : 表面；外表 synapse : 突触 irk : 惹恼；厌烦 escort :护送 escort :远征 探险 lousy :讨厌的

URLs for the blogs and papers to be read Interesting materials NN history2 Spheral CNN extreme learned image compression differential plasticity neural autoregressive flow Paper weekly clone voice (startup company github) soccer and ML weekly supervised learning ( previous work ) loc2vec Large scale training implementation distribution training Reinforcement learning reinforcement learning atari https://eng.uber.com/accelerated-neuroevolution/ Semantic segmentation Introduction Medical imaging blog Interesting blog faster rcnn grid learning course material max welling attention model (CVPR 2017) cvpr 2017 list image caption graph lstm reweight samples speech recognition reddit speech recognition summary good paper for speed speech recognition word2vec Deep RL examples REG Embedding speaker embedding automonous driving SHELL thirteen skills for writting shell script

Network description H1: 768(8*8*12) units, 1068 parameters(768+25*12)( bias associated with units, not kernel )

Vegetable leek : 韭菜 caraway; coriander : 香菜 spinach :菠菜 Chinese cabbage : 白菜 rape;cole :油菜 bitter gourd :苦瓜 white gourd :冬瓜 squash :菜瓜 string bean :四季豆 soybean :大豆；黄豆 mung bean :绿豆 bean sprout : 豆芽 lotus root : 莲藕 lotus seed :莲子 garlic : 蒜 green pepper :青椒 hot pepper :辣椒 ginger :生姜 scallion :青葱 taro :芋头 bamboo sprout : 竹笋 champignon :香菇 needle mushroom :金针菇 aloe :芦荟 artichoke : 洋蓟 asparagus : 芦笋；龙须菜；天冬 beetroot :甜菜根 bell pepper : 灯笼椒 broccoli :西兰花 brussels sprout :小羊白菜 cabbage :卷心菜 洋白菜 carrot :胡萝卜 cauliflower :花菜 celery :芹菜 corn :玉米 cucumber : 黄瓜 eggplant : 茄子 green bean :四季豆 lettuce : 生菜 mushroom :蘑菇 onion :洋葱 pea :豌豆 potato :土豆 pumpkin :南瓜 radish :小萝卜 sweet potato :红薯 tomato :西红柿 zucchini :绿皮西葫芦（又名意大利青瓜） laver : 紫菜 lean meat : 瘦肉 speck : 肥肉 agarics : 木耳 radish : 萝卜 silver fungi : 银耳 tendon : 健子肉 pork joint : 肘子肉 lentil : 小扁豆 mannequin : 人体模特 Fruit pineapple : 菠萝 凤梨 watermelon : 西瓜 pa...

Abastract This paper mainly focused on solving the problem of image-translation with unpaired images. Image-to-image translation is a class of vision and graphics problems where the goal is to learn to the mapping between an input image and an output image using a training set of aligned image pairs. However, obtaining large amount of paired images could be quite expensive. The author proposed learn a translation from source domain \(X\) to target domain \(Y\) in the absence of paired images. The goal is to learn a map \(G:X\rightarrow Y\), such that the distribution of \(G(X)\) is indistinguishable from \(Y\) using an adversarial loss. The author coupled it with reverse mapping \(F:Y\rightarrow X\) and cycle-consistency loss \(F(G(X))\approx X\) Fig.1: Given any two unordered image collections \(X\) and \(Y\), the algorithm could automatically "translate" an image from one into the other and vice versa: (left) 1074 Monet paintings and 6753 landscape photos fro...