µö·¯´× ±âÃÊ À̷кÎÅÍ ANN, ¿ÀÅäÀÎÄÚ´õ, CNN, RNN, GAN, FCN, DQN, À̹ÌÁö ĸ¼Å´× ÃֽŠ¸ðµ¨ ±¸Çö±îÁö ¡ºÅÙ¼Ç÷ηΠ¹è¿ì´Â µö·¯´×¡»¿¡¼´Â µö·¯´× ±â¹ýÀÇ ÀÌ·ÐÀû ¹è°æÀÌ µÇ´Â ±âÃÊÀûÀÎ ¼öÇÐÀû À̷еéÀ» ÀÚ¼¼ÇÏ°Ô ¼Ò°³ÇÏ°í, µö·¯´× ±âÃÊ ¸ðµ¨µé(ANN, ¿ÀÅäÀÎÄÚ´õ, CNN, RNN)ÀÇ Á¤È®ÇÑ ÀÌÇظ¦ À§ÇØ ÅÙ¼Ç÷Π¿¹Á¦ ÄÚµå¿Í ÇÔ²² ¼³¸íÇÑ´Ù. ¶ÇÇÑ, µö·¯´× ¸ðµ¨µéÀ» ´Ù¾çÇÑ ¹®Á¦¿¡ Àû¿ëÇÏ°í ½ÇÁ¦ ¹®Á¦¿¡ ÀÀ¿ëÇÏ´Â ¹æ¹ýÀ» ¼Ò°³ÇÑ´Ù. Ã¥ÀÇ Ãʹݿ¡´Â ¼±Çü ´ë¼ö, È®·ü Åë°è, ÃÖÀûÈ À̷аú °°Àº ¼öÇÐÀû ÀÌ·ÐÀ» ¼³¸íÇÏ°í, µö·¯´× ¾Ë°í¸®ÁòÀÇ ±âº» ±¸Á¶ÀÎ ANN, ¿ÀÅäÀÎÄÚ´õ, CNN, RNNÀ» ´Ù·é´Ù. Á߹ݿ¡´Â ¾Õ¿¡¼ ¹è¿î ANN, CNN, RNN ±¸Á¶¸¦ À̹ÌÁö ĸ¼Å´×, Semantic Image Segmentation ¹®Á¦¿¡ ¾î¶»°Ô ÀÀ¿ëÇÏ´ÂÁö¸¦ ¼³¸íÇÑ´Ù. Ã¥ÀÇ ÈĹݿ¡´Â ÃÖ±Ù¿¡ Àαâ ÀÖ´Â ÁÖÁ¦ÀÎ »ý¼º ¸ðµ¨°ú °È ÇнÀÀÇ °³³äÀ» »ìÆ캸°í, ÆÄÀÎ Æ©´×°ú »çÀü ÇнÀµÈ ¸ðµ¨À» ÀÌ¿ëÇؼ ½ÇÁ¦ ¹®Á¦¸¦ ÇØ°áÇÏ´Â ¹æ¹ýÀ» ¹è¿î´Ù. 1±ÇÀÇ Ã¥À¸·Î µö·¯´× ±âÃÊ À̷кÎÅÍ ÅÙ¼Ç÷Π¶óÀ̺귯¸®¸¦ ÀÌ¿ëÇÑ ½ÇÁ¦ ±¸Çö±îÁö ¸ðµÎ ÆľÇÇÒ ¼ö ÀÖ´Ù.
¼¿ï´ëÇб³ ÀΰøÁö´É ¹× ÄÄÇ»ÅÍ ºñÀü ¿¬±¸½Ç¿¡¼ ¼®»çÇÐÀ§¸¦ ¹Þ¾Ò´Ù. ÅÙ¼ÇÃ·Î¿Í ÀΰøÁö´É, ¸Ó½Å·¯´×, µö·¯´×À» °ü·Ã ³»¿ëÀ» Æ÷½ºÆÃÇÏ´Â ¡°¼Ö¶ó¸®½ºÀÇ ÀΰøÁö´É ¿¬±¸½Ç¡±(solarisailab.com)À̶ó´Â ºí·Î±×¸¦ ¿î¿µ ÁßÀÌ´Ù.
1. ÀΰøÁö´É, ¸Ó½Å·¯´×, µö·¯´× ¼Ò°³1.1 µö·¯´× ¾Ë°í¸®ÁòÀÇ µîÀå¹è°æ1.2 Áöµµ ÇнÀ1.3 ºñÁöµµ ÇнÀ1.4 °È ÇнÀ1.5 Á¤¸®2. ÅÙ¼Ç÷Π¼Ò°³2.1 ÅÙ¼Ç÷Π¼³Ä¡ ¹× Ã¥¿¡¼ »ç¿ëÇÏ´Â ¼Ò½º ÄÚµå ´Ù¿î·Îµå2.1.1 ÅÙ¼Ç÷Π¼Ò°³ 2.1.2 ÅÙ¼Ç÷Π¼³Ä¡ 2.1.3 Ã¥¿¡¼ »ç¿ëÇÏ´Â ¼Ò½º ÄÚµå ´Ù¿î·Îµå 2.2 µö·¯´×, ÅÙ¼Ç÷ΠÀÀ¿ë ºÐ¾ß2.2.1 ÄÄÇ»ÅÍ ºñÀü 2.2.2 ÀÚ¿¬¾î ó¸®2.2.3 À½¼º ÀÎ½Ä 2.2.4 °ÔÀÓ 2.2.5 »ý¼º ¸ðµ¨ 2.3 ÅÙ¼Ç÷ΠÃß»óÈ ¶óÀ̺귯¸®µé2.3.1 ÄÉ¶ó½º 2.3.2 TF-Slim2.3.3 Sonnet2.4 Á¤¸®3. ÅÙ¼Ç÷Π±âÃÊ¿Í ÅÙ¼º¸µå3.1 ÅÙ¼Ç÷Π±âÃÊ - ±×·¡ÇÁ »ý¼º°ú ±×·¡ÇÁ ½ÇÇà3.2 Ç÷¹À̽ºÈ¦´õ3.3 ¼±Çüȸ±Í ¹× °æ»çÇÏ°¹ý ¾Ë°í¸®Áò3.3.1 ¸Ó½Å·¯´×ÀÇ ±âº» ÇÁ·Î¼¼½º - °¡¼³ Á¤ÀÇ, ¼Õ½Ç ÇÔ¼ö Á¤ÀÇ, ÃÖÀûÈ Á¤ÀÇ 3.3.2 ¼±Çü ȸ±Í ¾Ë°í¸®Áò ±¸Çö ¹× º¯¼ö3.4 ÅÙ¼º¸µå¸¦ ÀÌ¿ëÇÑ ±×·¡ÇÁ ½Ã°¢È3.5 Á¤¸®4. ¸Ó½Å·¯´× ±âÃÊ À̷еé4.1 Batch Gradient Descent, Mini-Batch Gradient Descent, Stochastic Gradient Descent4.2 Training Data, Validation Data, Test Data ¹× ¿À¹öÇÇÆÃ4.3 ¼ÒÇÁÆ®¸Æ½º ȸ±Í4.3.1 ¼ÒÇÁÆ®¸Æ½º ȸ±Í 4.3.2 Å©·Î½º ¿£Æ®·ÎÇÇ ¼Õ½Ç ÇÔ¼ö 4.3.3 MNIST µ¥ÀÌÅͼ 4.3.4 One-hot Encoding 4.4 ¼ÒÇÁÆ®¸Æ½º ȸ±Í¸¦ ÀÌ¿ëÇÑ MNIST ¼ýÀÚ ºÐ·ù±â ±¸Çö4.4.1 mnist_classification_using_softmax_regression.py 4.4.2 tf_nn_sparse_softmax_cross_entropy_with_logits_example.py 4.5 Á¤¸®5. Àΰø½Å°æ¸Á(ANN)5.1 Àΰø½Å°æ¸ÁÀÇ µîÀå ¹è°æ5.2 ÆÛ¼ÁÆ®·Ð5.3 ´ÙÃþÆÛ¼ÁÆ®·Ð MLP5.4 ¿À·ù¿ªÀüÆÄ ¾Ë°í¸®Áò5.5 ANNÀ» ÀÌ¿ëÇÑ MNIST ¼ýÀÚ ºÐ·ù±â ±¸Çö5.6 Á¤¸®6. ¿ÀÅäÀÎÄÚ´õ(AutoEncoder)6.1 ¿ÀÅäÀÎÄÚ´õÀÇ °³³ä6.2 ¿ÀÅäÀÎÄÚ´õ¸¦ ÀÌ¿ëÇÑ MNIST µ¥ÀÌÅÍ À籸Ãà6.3 ¿ÀÅäÀÎÄÚ´õ¿Í ¼ÒÇÁÆ®¸Æ½º ºÐ·ù±â¸¦ ÀÌ¿ëÇÑ MNIST ºÐ·ù±â ±¸Çö6.3.1 ÆÄÀÎ Æ©´×°ú ÀüÀÌ ÇнÀ 6.3.2 ¿ÀÅäÀÎÄÚ´õ¿Í ¼ÒÇÁÆ®¸Æ½º ºÐ·ù±â¸¦ ÀÌ¿ëÇÑ MNIST ¼ýÀÚ ºÐ·ù±â ±¸Çö6.4 Á¤¸®7. ÄÁº¼·ç¼Ç ½Å°æ¸Á(CNN)7.1 ÄÁº¼·ç¼Ç ½Å°æ¸ÁÀÇ °³³ä - ÄÁº¼·ç¼Ç, Ç®¸µ7.2 MNIST ¼ýÀÚ ºÐ·ù¸¦ À§ÇÑ CNN ºÐ·ù±â ±¸Çö7.3 CNNÀ» ÀÌ¿ëÇÑ CIFAR-10 À̹ÌÁö ºÐ·ù±â ±¸Çö7.3.1 CIFAR-10 µ¥ÀÌÅͼÂ7.3.2 µå·Ó¾Æ¿ô 7.3.3 CNNÀ» ÀÌ¿ëÇÑ CIFAR-10 À̹ÌÁö ºÐ·ù±â ±¸Çö 7.4 ´ëÇ¥ÀûÀÎ CNN ¸ðµ¨µé - AlexNet, VGGNet, GoogLeNet, ResNet7.4.1 AlexNet 7.4.2 VGGNet 7.4.3 GoogLeNet(Inception v1) 7.4.4 ResNet 7.5 tf.train.Saver API¸¦ ÀÌ¿ëÇؼ ¸ðµ¨°ú ÆĶó¹ÌÅ͸¦ ÀúÀåÇÏ°í ºÒ·¯¿À±â7.6 Á¤¸®8. ¼øȯ½Å°æ¸Á(RNN)8.1 ¼øȯ½Å°æ¸Á8.2 LSTM(Àå/´Ü±â ±â¾ï ³×Æ®¿öÅ©)¿Í °æ»çµµ »ç¶óÁü ¹®Á¦8.3 GRU8.4 ÀÓº£µù8.4.1 ÀÓº£µùÀÇ°³³ä8.4.2 tf.nn.embedding_lookupÀ» ÀÌ¿ëÇÑ ÀÓº£µù ±¸Çö 8.5 °æ»çµµ Áõ°¡ ¹®Á¦¿Í °æ»çµµ ÀÚ¸£±â8.6 Char-RNN8.6.1 Char-RNNÀÇ °³³ä 8.6.2 ÅÙ¼Ç÷θ¦ ÀÌ¿ëÇÑ Char-RNN ±¸Çö 8.6.2.1 train_and_sampling.py 8.6.2.2 utils.py 8.7 Á¤¸®9. À̹ÌÁö ĸ¼Å´×(Image Captioning)9.1 À̹ÌÁö ĸ¼Å´× ¹®Á¦ ¼Ò°³9.2 À̹ÌÁö ĸ¼Å´× µ¥ÀÌÅͼ - MS COCO9.3 À̹ÌÁö ĸ¼Å´× ±¸Çö - im2txt9.4 im2txt ÄÚµå ±¸Á¶¿¡ ´ëÇÑ ¼³¸í ¹× ÄÚµå ½ÇÇà ¹æ¹ý9.4.1 train.py 9.4.2 show_and_tell_model.py 9.4.3 run_inference.py 9.5 Á¤¸®10. Semantic Image Segmentation10.1 Semantic Image Segmentation °³³ä10.2 FCN10.3 Semantic Image SegmentationÀ» À§ÇÑ µ¥ÀÌÅͼ - MIT Scene Parsing10.4 FCNÀ» ÀÌ¿ëÇÑ Semantic Image Segmentation ±¸Çö - FCN.tensorflow10.4.1 FCN.py 10.4.2 TensorflowUtils.py 10.4.3 read_MITSceneParsingData.py 10.4.4 BatchDatsetReader.py10.5 Á¤¸®11. »ý¼º ¸ðµ¨ - GAN11.1 »ý¼º ¸ðµ¨ÀÇ °³³ä11.2 GANÀÇ °³³ä11.3 GANÀ» ÀÌ¿ëÇÑ MNIST µ¥ÀÌÅÍ »ý¼º11.4 Á¤¸®12. °È ÇнÀ(Reinforcement Learning)12.1 °È ÇнÀÀÇ ±âº» °³³ä°ú MDP12.1.1 »óÅ °¡Ä¡ ÇÔ¼ö 12.1.2 Çൿ °¡Ä¡ ÇÔ¼ö 12.2 Q-Learning12.2.1 Q-Table°ú Q-Networks 12.2.2 ¡ô-Greedy 12.3 DQN12.4 DQNÀ» ÀÌ¿ëÇÑ °ÔÀÓ ¿¡ÀÌÀüÆ® ±¸Çö - CatchGame12.4.1 train_catch_game.py 12.4.2 play_catch_game.ipynb 12.5 Á¤¸®13. ÆÄÀÎ Æ©´×°ú »çÀü ÇнÀµÈ ¸ðµ¨À» ÀÌ¿ëÇÑ ½ÇÁ¦ ¹®Á¦ ÇØ°á13.1 ÆÄÀÎ Æ©´× ¹× ÀüÀÌ ÇнÀ ±â¹ý ¸®ºä13.2 Inception v3 RetrainingÀ» ÀÌ¿ëÇÑ ³ª¸¸ÀÇ ºÐ·ù±â13.2.1 Inception v3 ¸ðµ¨ 13.2.2 inceptionv3_retrain.py - ³ª¸¸ÀÇ µ¥ÀÌÅͼÂÀ¸·Î ÆÄÀÎ Æ©´×13.2.3 inceptionv3_retrain.py 13.2.4 inceptionv3_inference.py 13.3 »çÀü ÇнÀµÈ ¸ðµ¨À» ÀÌ¿ëÇÑ ¹°Ã¼ °ËÃâ ¼öÇà13.3.1 ¹°Ã¼ °ËÃâÀÇ °³³ä 13.3.2 »çÀü ÇнÀµÈ Faster R-CNN ¸ðµ¨·Î ¹°Ã¼ °ËÃâ ¼öÇà13.3.3 faster_rcnn_inference.py 13.4 TensorFlow Hub13.5 Á¤¸®13.6 ´õ °øºÎÇÒ °Íµé