{"id":28472,"date":"2020-03-26T10:52:47","date_gmt":"2020-03-26T01:52:47","guid":{"rendered":"https:\/\/smallake.kr\/?p=28472"},"modified":"2020-03-26T10:52:47","modified_gmt":"2020-03-26T01:52:47","slug":"todotrader%ec%9d%98-deeptrading-with-tensorflow","status":"publish","type":"post","link":"https:\/\/smallake.kr\/?p=28472","title":{"rendered":"TodoTrader\uc758 DeepTrading with TensorFlow"},"content":{"rendered":"

1.
\n\ud2b8\ub808\uc774\ub529\uacfc \uad00\ub828\ud55c \uae00\uc744 \uc4f0\ub294 \ube14\ub7ec\uac70\ub4e4\ub3c4 \uc2dc\ub958\uc5d0 \ub530\ub77c \ubd80\uce68\uc744 \uacaa\uc2b5\ub2c8\ub2e4. HFT\uac00 \uac01\uad11\uc744 \ubc1b\uc558\ub358 \ub54c \uc790\uc8fc \ucc3e\uc558\ub358 \uacf3\ub4e4\uc911 \uc9c0\uae08\uae4c\uc9c0 \ub0a8\uc544 \uc788\ub294 \uacf3\uc740 \uac70\uc758 \uc5c6\uc2b5\ub2c8\ub2e4. \ub9dd\ud558\uc9c0 \uc54a\uc744 \uac83\uac19\uc558\ub358 Lowlatency.com\ub3c4 \uba85\ub9e5\uc774 \ub04a\uc5b4\uc84c\uc2b5\ub2c8\ub2e4. \uc694\uc998 \ub9ce\uc774 \uc62c\ub77c\uc624\ub294 \uae00\ub4e4\uc740 \uae30\uacc4\ud559\uc2b5\uacfc \uad00\ub828\ud55c \uacf3\ub4e4\uc774 \uc788\uc2b5\ub2c8\ub2e4. \uadf8\uc911 \ub450\uac1c\ub97c \uc18c\uac1c\ud569\ub2c8\ub2e4. \uccab\uc9f8 Robot Wealth<\/a>\uc785\ub2c8\ub2e4. R\uc744 \uc774\uc6a9\ud55c \uc0ac\ub840\ub97c \uc18c\uac1c\ud558\uace0 \uc788\uace0 Zerro Platform\uc744 \uc774\uc6a9\ud558\uace0 \uc788\uc2b5\ub2c8\ub2e4. \ub3c5\uc77c \uc2a4\ud0c0\ud2b8\uc5c5\uc73c\ub85c \uc774\ud574\ud558\ub294 Zorro Platform\uc740 C\ubb38\ubc95\uc744 \uc774\uc6a9\ud558\uc5ec \uc804\ub7b5\uc744 \uac1c\ubc1c\ud560 \uc218 \uc788\ub3c4\ub85d \ud558\ub294 \ud658\uacbd\uc744 \uc81c\uacf5\ud569\ub2c8\ub2e4.<\/p>\n

Deep Learning for Trading Part 1: Can it Work?<\/a>
\nDeep Learning for Trading Part 2: Configuring TensorFlow and Keras to run on GPU<\/a>
\nDeep Learning for Trading Part 3: Feed Forward Networks<\/a><\/p>\n

Machine learning for Trading:What Does it Really Look Like?<\/a>
\nMachine learning for Trading: Part 2<\/a><\/p>\n

\ub458\uc9f8\ub294 TodoTrader\uc785\ub2c8\ub2e4. \uc55e\uc11c Robot Wealth\uac00 R\uc5b8\uc5b4\ub97c \uc8fc\ub85c \uc0ac\uc6a9\ud55c \uac83\uc5d0 \ube44\ud558\uc5ec Python\uacfc Tensorflow\ub97c \uc774\uc6a9\ud55c \ud2b8\ub808\uc774\ub529\uc2dc\uc2a4\ud15c\uc744 \uc8fc\ub85c \ub2e4\ub8e8\uace0 \uc788\uc2b5\ub2c8\ub2e4. \uc5ec\uae30\ub3c4 \uc544\uc27d\uac8c 19\ub144\ub9d0\uae4c\uc9c0 \uae00\uc744 \uc62c\ub77c\uc624\ub2e4\uac00 \ucd5c\uadfc\uc5d0\ub294 \ub738\ud569\ub2c8\ub2e4. TodoTrader\ub97c \uc6b4\uc601\ud558\ub294 \uc0ac\ub78c\uc740 Ramon Martin<\/a>\uc778\ub370 \uae30\uacc4\ud559\uc2b5(Python)\uacfc \uad00\ub828\ud55c \uad50\uc721\uc774\ub098 \ucee8\uc124\ud305\uc744 \ud558\ub294 \ub4ef \ud569\ub2c8\ub2e4. \uc62c\ub77c\uc628 \uae00\ub4e4\uc785\ub2c8\ub2e4.<\/p>\n

\uba3c\uc800 DeepTrading with TensorFlow<\/a>\uc740 Tensorflow\uc758 \uae30\ubcf8\uc744 \ub2e4\ub8f9\ub2c8\ub2e4. Tensorflow\uac00 \ub3d9\uc791\ud558\ub294 \ud750\ub984\uc5d0 \ub530\ub77c \uac04\ub7b5\ud55c \uc124\uba85\uc744 \uc81c\uacf5\ud569\ub2c8\ub2e4.<\/p>\n

\"\"<\/p>\n

\ub450\ubc88\uc9f8 DeepTrading with TensorFlow II<\/a>\uc758 \uc8fc\uc81c\ub294 ‘How to create elemental NN(Neural Network) tensors in TensorFlow’\uc785\ub2c8\ub2e4.
\n\uc138\ubc88\uc9f8 DeepTrading with TensorFlow III<\/a>\uc758 \uc8fc\uc81c\ub294 ‘Implementing a one hidden layer Neural Network’\uc785\ub2c8\ub2e4.<\/p>\n

\"\"<\/p>\n

\ub124\ubc88\uc9f8 DeepTrading with Tensorflow IV<\/a>\uc758 \uc8fc\uc81c\ub294 ‘Implementing a one hidden layer Neural Network with save and restore’\uc785\ub2c8\ub2e4. \ubaa8\ub378\uc758 \uc7ac\uc0ac\uc6a9\uc774 \uac00\ub2a5\ud55c Save\uc640 Restore\ub97c \ub367\ubd99\uc778 \ubaa8\ud615\uc785\ub2c8\ub2e4.<\/p>\n

The progress of the model can be saved during and after training. This means that a model can be resumed where it left off and avoid long training times. Saving also means that you can share your model and others can recreate your work.<\/p><\/blockquote>\n

\ub2e4\uc12f\ubc88\uc9f8 Deep Trading with TensorFlow V<\/a>\uc758 \uc8fc\uc81c\ub294 ‘Implementing a multiple hidden layer Neural Network’\uc785\ub2c8\ub2e4. \uc55e\uc11c TensorFlow III\uac00 One Hidden Layer\uacfc \ube44\uad50\ud574\ubcf4\uc2dc\uba74 \ub429\ub2c8\ub2e4.
\n\"\"<\/p>\n

\uc5ec\uc12f\ubc88\uc9f8\ubd80\ud130\ub294 \uc55e\uc11c\uc758 \ubaa8\ud615\uc744 \uc2e4\uc81c \uae08\uc735\uc2dc\uc7a5\uc5d0 \uc801\uc6a9\ud558\ub294 \uc0ac\ub840\ub4e4\uc785\ub2c8\ub2e4. \uadf8\uc911 \uccab\ubc88\uc9f8\ub294 DeepTrading with TensorFlow VI<\/a>\ub85c \uc8fc\uc81c\ub294 ‘Making the first prediction in the Forex Market’\uc785\ub2c8\ub2e4. \ub450\ubc88\uc9f8 \uc2dc\uc7a5\uc740 \uae08\uc2dc\uc7a5\uc785\ub2c8\ub2e4. \uc77c\uacf1\ubc88\uc9f8 DeepTrading with TensorFlow VII<\/a>\ub85c \uc8fc\uc81c\ub294 ‘Making prediction of GOLD price’\ub85c \uae08\uc2dc\uc7a5\uc785\ub2c8\ub2e4.<\/p>\n

\uc774\uc0c1\uc758 \uc8fc\uc81c\uc640 \uad00\ub828\ud55c Jupyter notebook\uc6a9 \ud30c\uc77c\ub4e4\uc740 \uc544\ub798\uc5d0\uc11c \ud655\uc778\ud558\uc2e4 \uc218 \uc788\uc2b5\ub2c8\ub2e4. \uc774\uc0c1\uc758 \uc5f0\uc7ac\ubb3c\uacfc \uad00\ub828\ud55c \ud504\ub85c\uc81d\ud2b8 \uc18c\uc2a4\ub294 DeepTrading<\/a>\ub97c \ucc38\uc870\ud558\uc138\uc694.<\/p>\n

01_How_TensorFlow_Works.ipynb<\/a>
\n02_Tensors.ipynb<\/a>
\n03_One_Hidden_Layer_NN.ipynb<\/a>
\n04_One_Hidden_Layer_NN_Save_Restore.ipynb<\/a><\/a>
\n05_Multiple_Hidden_Layers_NN.ipynb<\/a><\/a>
\n06_Multiple_Hidden_Layers_NN-Restoring_Model.ipynb<\/a>
\n07_First_Forex_Prediction-240Min.ipynb<\/a>
\n07_First_Forex_Prediction-60Min.ipynb<\/a>
\n07_First_Forex_Prediction_15Min.ipynb<\/a>
\n08_Gold_Prediction-60Min.ipynb<\/a><\/p>\n

3.
\n\uc774\uc81c \ud559\uc2b5\uc73c\ub85c \ud68d\ub4dd\ud55c \uc9c0\uc2dd\uc744 \uc2e4\uac70\ub798\uc2dc\uc2a4\ud15c\uc5d0 \uc801\uc6a9\ud560 \ucc28\ub840\uc785\ub2c8\ub2e4. \uc5ec\ub35f\ubc88\uc9f8 Deep Trading with TensorFlow VIII<\/a>\uc740 Deployment\uc640 \uad00\ub828\ud55c \uc8fc\uc81c\ub97c \ub2e4\ub85d \uc788\uc2b5\ub2c8\ub2e4. \uc81c\ubaa9\uc740 ‘Trading system Tensorflow serving’\uc785\ub2c8\ub2e4.<\/p>\n

\"\"<\/p>\n

After training and testing your model, you had a great question: how will you deploy the model and make predictions for new data samples? Luckily for us, TensorFlow was developed for production, and it provides a robust solution for model deployment, known as TensorFlow Serving<\/a>. In this approaching, we have three steps:<\/p>\n