随笔分类 -  [39] 模式识别/机器学习/优化

摘要:https://zhuanlan.zhihu.com/p/64361703 https://github.com/ChristosChristofidis/awesome-deep-learning Table of Contents Books Courses Videos and Lecture 阅读全文
posted @ 2021-06-03 23:27 emanlee 阅读(190) 评论(0) 推荐(0)
摘要:计算机视觉 Computer Vision 会议: AAAI: AAAI Conference on Artificial IntelligenceACCV: Asian Conference on Computer VisionACM MM: ACM International Conferenc 阅读全文
posted @ 2021-06-03 19:25 emanlee 阅读(921) 评论(0) 推荐(0)
摘要:最近深度学习两巨头 Bengio 和 LeCun 在 ICLR 2020 上点名 Self-Supervised Learning(SSL,自监督学习) 是 AI 的未来,而其的代表的 Framework 便是 Contrastive Learning(CL,对比学习)。 另一巨头 Hinton 和 阅读全文
posted @ 2021-06-01 19:36 emanlee 阅读(1056) 评论(0) 推荐(0)
摘要:2021-06-01 10:32:15.183997: W tensorflow/python/util/util.cc:348] Sets are not currently considered sequences, but this may change in the future, so c 阅读全文
posted @ 2021-06-01 15:56 emanlee 阅读(694) 评论(0) 推荐(0)
摘要:from keras.utils import to_categorical module 'tensorflow.compat.v2' has no attribute '__internal__ from keras.utils import to_categorical $ python te 阅读全文
posted @ 2021-05-31 23:06 emanlee 阅读(5418) 评论(0) 推荐(0)
摘要:验证集loss上升,准确率却上升 验证集loss上升,acc也上升这种现象很常见,原因是过拟合或者训练验证数据分布不一致导致,即在训练后期,预测的结果趋向于极端,使少数预测错的样本主导了loss,但同时少数样本不影响整体的验证acc情况。ICML2020发表了一篇文章:《Do We Need Zer 阅读全文
posted @ 2021-05-26 22:51 emanlee 阅读(9482) 评论(0) 推荐(1)
摘要:确保 tensorflow 成功安装 conda install numpy pandas jupyter notebook matplotlibconda install scipy h5pypip install TFLearn 阅读全文
posted @ 2021-03-07 22:07 emanlee 阅读(162) 评论(0) 推荐(0)
摘要:import pydotimport graphviz from keras.utils import plot_model plot_model(model, to_file='model.png') ('Failed to import pydot. You must `pip install 阅读全文
posted @ 2021-03-05 08:32 emanlee 阅读(3518) 评论(0) 推荐(1)
摘要:R Run as Administrator: install.packages('devtools') install.packages("fansi") library('devtools') install_github("MRCIEU/ieugwasr") install_github("W 阅读全文
posted @ 2021-03-03 16:16 emanlee 阅读(1579) 评论(0) 推荐(0)
摘要:Building wheels for collected packages: fbprophet Building wheel for fbprophet (setup.py) ... error ERROR: Command errored out with exit status 1: com 阅读全文
posted @ 2021-03-02 21:08 emanlee 阅读(1950) 评论(0) 推荐(0)
摘要:.c(49): warning: #1-D: last line of file ends without a newline 最后一行回车后还要保证是在第一列,新一行不能有空格等空字符才能不出此警告 这是KEIL的bug: 需要代码文件的最后两行为空行,并且没有空格。 阅读全文
posted @ 2021-03-01 16:49 emanlee 阅读(2044) 评论(0) 推荐(0)
摘要:attention_vector = np.mean(get_activations(m, testing_inputs_1, print_shape_only=True, layer_name='attention_vec')[0], axis=2).squeeze()funcs = [K.fun 阅读全文
posted @ 2021-02-27 10:47 emanlee 阅读(1593) 评论(2) 推荐(0)
摘要:tensorflow.python.framework.errors_impl.UnknownError: Fail to find the dnn implementation. [[{{node CudnnRNN}}]] [[model/lstm/PartitionedCall]] [Op:__ 阅读全文
posted @ 2021-02-26 23:39 emanlee 阅读(1356) 评论(0) 推荐(0)
摘要:TypeError: ('Keyword argument not understood:', 'input') model = Model(input=[inputs], output=output) 报错信息TypeError: ('Keyword argument not understood 阅读全文
posted @ 2021-02-26 23:38 emanlee 阅读(5330) 评论(0) 推荐(1)
摘要:旧版本中: from keras.layers import merge merge6 = merge([layer1,layer2], mode = 'concat', concat_axis = 3)新版本中: from keras.layers.merge import concatenate 阅读全文
posted @ 2021-02-26 23:32 emanlee 阅读(307) 评论(0) 推荐(0)
摘要:打开CMD 命令行窗口,输入 nvcc --version nvcc --versionnvcc: NVIDIA (R) Cuda compiler driverCopyright (c) 2005-2020 NVIDIA CorporationBuilt on Wed_Jul_22_19:09:3 阅读全文
posted @ 2021-02-26 23:11 emanlee 阅读(1959) 评论(0) 推荐(0)
摘要:2021-02-26 22:54:13.146272: I tensorflow/core/common_runtime/gpu/gpu_device.cc:1406] Created TensorFlow device (/job:localhost/replica:0/task:0/device 阅读全文
posted @ 2021-02-26 22:58 emanlee 阅读(3443) 评论(0) 推荐(0)
摘要:conda install keras 阅读全文
posted @ 2021-02-26 22:09 emanlee 阅读(127) 评论(0) 推荐(0)
摘要:change: moving_avg = pd.rolling_mean(ts_log,12) to: moving_avg = ts_log.rolling(12).mean() 阅读全文
posted @ 2021-02-22 09:48 emanlee 阅读(1020) 评论(0) 推荐(0)
摘要:model.fit(trainX, trainY, nb_epoch=200, batch_size=2, verbose=2) It's just epochs now. nb_epoch was deprecated years ago. 把 nb_epoch 修改 为 epochs 阅读全文
posted @ 2021-02-22 09:25 emanlee 阅读(4757) 评论(0) 推荐(0)