ÆÄÀ̽ãÀº ¼±ÅÃÀÌ ¾Æ´Ñ Çʼö! ÆÄÀÌÅäÄ¡·Î µö·¯´× ÀÔ¹®Çϱâ!µö·¯´× ±â¼ú¿¡ ÃÊÁ¡À» µÎ°í µö·¯´×À» ±¸ÇöÇϱâ À§ÇØ ÆÄÀ̽ãÀ» ÀÌ¿ëÇϴµ¥, ¸Ó½Å·¯´× ¶óÀ̺귯¸®ÀÎ ÆÄÀÌÅäÄ¡¸¦ È°¿ëÇÏ¿© ´Ù¾çÇÑ ÅÙ¼¸¦ Áö¿øÇÏ´Â ¹æ¹ýÀ» ¾Ë¾Æº»´Ù. ÆÄÀÌÅäÄ¡´Â ÆÄÀ̽ã ÄÚµù°ú ºñ½ÁÇϱ⠶§¹®¿¡ ¾ð¾î°¡ ¾î·ÆÁö ¾Ê´Ù. Äڵ尡 °£°áÇÏ°í ³À̵µ°¡ ³·¾Æ ÅÙ¼Ç÷ο캸´Ù »ç¿ëÇϱâ ÈξÀ ½±´Ù´Â Ư¡ÀÌ ÀÖ´Ù. ÇÁ·Î±×·¡¹Ö ¾ð¾îÀÇ ±âº»Àû ¼öÁظ¸ °®Ãß°í ÀÖ´Ù¸é °í±Þ ½ºÅ³ÀÌ ¾ø¾îµµ Äڵ带 ÀÛ¼ºÇغ¸¸ç Á÷Á¢ ½ÇÇàÇغ¼ ¼ö ÀÖµµ·Ï ±¸¼ºÇÏ¿´±â ¶§¹®¿¡ Àǹ̸¦ Á¤È®ÇÏ°í °³³äÀ» ÀÌÇØÇÒ ¼ö ÀÖ´Ù. ÇнÀÀ» ½ÃÀÛÇϱâ Àü ±âº»ÀûÀÎ ³»¿ë°ú ÄÚµå ÀÛ¼ºÀ» À§ÇÑ ½Ã½ºÅÛ È¯°æ ±¸ÃàºÎÅÍ ½ÃÀÛÇÏ¿©, ¿äÁò ½±°Ô µé¸®´Â ¸Ó½Å·¯´×, µö·¯´×, ÀΰøÁö´É µîÀÇ °³³äÀ» ½±°Ô ¼³¸íÇÏ°í È°¿ë ºÐ¾ßµµ ¾Ë¾Æº»´Ù. ƯÈ÷ Áß°£ Áß°£ ¿¹Á¦¸¦ ¼ö·ÏÇÏ¿© ÄÚµå¿Í ¼³¸íÀ» ÀÚ¼¼ÇÏ°Ô ¼³¸íÇϱ⠶§¹®¿¡ Ãʺ¸Àڵ鵵 ½±°Ô Á¢±ÙÇÒ ¼ö ÀÖ´Ù´Â ÀåÁ¡ÀÌ ÀÖ´Ù. Äڵ带 µû¶óÇϱâ À§ÇÑ ½Ç½À ÆÄÀÏ ´Ù¿î·Îµå´Â Á¤º¸¹®È»ç ȨÆäÀÌÁö(infopub.co.kr) ÀÚ·á½Ç¿¡¼ °¡´ÉÇϸç, ÇнÀ Áß ±Ã±ÝÇÑ »çÇ×Àº ÀúÀÚÀÇ github.com/Justin-A/DeepLearning101/issues¿¡¼ Çǵå¹é °¡´ÉÇÏ´Ù.
¼º±Õ°ü´ëÇб³ Åë°èÇаú¸¦ Àü°øÇÏ°í, ºòµ¥ÀÌÅÍ ¿¬ÇÕ ÇÐȸ Åõºò½º¸¦ ¸¸µé¾ú´Ù. ÇöÀç ¿¬¼¼´ëÇб³ »ê¾÷°øÇаú¿¡¼ ¹Ú»ç°úÁ¤À» ¹â°í ÀÖ´Ù. µ¥ÀÌÅÍ ºÐ¼® ¹× ÀΰøÁö´É °ü·Ã ´ëȸ¿¡¼ ´Ù¼öÀÇ ¼ö»óÀ» ÇÏ¿´À¸¸ç µ¥ÀÌÅÍ ºÐ¼®°ú ÀΰøÁö´É Àü ºÐ¾ß¿¡ °ü½ÉÀ» °®°í ¿¬±¸ ÁßÀÌ´Ù. ƯÈ÷ µö·¯´×°ú °ÈÇнÀ¿¡ °ü½ÉÀ» °¡Áö°í ÀÖÀ¸¸ç, ºí·Î±× ¿î¿µ ¹× ´Ù¾çÇÑ °¿¬ È°µ¿ µîÀ» ÁøÇàÇÏ°í ÀÖ´Ù.
Part 01 ÆÄÀÌÅäÄ¡ ±âÃÊ1. ÆÄÀ̽㠶Ǵ ¾Æ³ªÄÜ´Ù ¼³Ä¡Çϱâ 1.1 ÆÄÀ̽㠰ø½Ä ȨÆäÀÌÁö¿¡¼ ´Ù¿î·ÎµåÇϱâ 1.2 ¾Æ³ªÄÜ´Ù¸¦ ÀÌ¿ëÇØ ÆÄÀ̽㠴ٿî·ÎµåÇϱâ 1.3 °ø½Ä ȨÆäÀÌÁö¿¡¼ ÆÄÀ̽㠼³Ä¡Çϱâ vs. ¾Æ³ªÄÜ´Ù¸¦ ÀÌ¿ëÇØ ÆÄÀ̽㠼³Ä¡Çϱâ 1.4 °¡»ó ȯ°æ ¼³Á¤Çϱâ 1.5 ÁÖÇÇÅÍ ³ëÆ®ºÏ ¼³Ä¡ ¹× ½ÇÇà 2. CUDA, CuDNN ¼³Ä¡Çϱâ 2.1 CPU vs. GPU 2.2 CUDA ¿ªÇÒ ¹× ¼³Ä¡Çϱâ 2.3 CuDNN ¿ªÇÒ ¹× ¼³Ä¡Çϱâ 2.4 Docker¶õ? 3. ÆÄÀÌÅäÄ¡ ¼³Ä¡Çϱâ 4. ¹Ýµå½Ã ¾Ë¾Æ¾ß ÇÏ´Â ÆÄÀÌÅäÄ¡ ½ºÅ³ 4.1 ÅÙ¼ 4.2 Autograd Part 02 AI Background1. ÀΰøÁö´É(µö·¯´×)ÀÇ Á¤ÀÇ¿Í »ç·Ê 1.1 ÀΰøÁö´ÉÀ̶õ? 1.2 ÀΰøÁö´ÉÀÇ »ç·Ê 2. ÆÄÀÌÅäÄ¡ 3. ¸Ó½Å·¯´×ÀÇ Á¤ÀÇ¿Í Á¾·ù 3.1 ¸Ó½Å·¯´×À̶õ? 3.2 ¸Ó½Å·¯´×ÀÇ Á¾·ù 3.3 ¸Ó½Å·¯´×ÀÇ ±¸ºÐ 3.4 ÁöµµÇнÀ ¸ðµ¨ÀÇ Á¾·ù 4. °úÀûÇÕ 4.1 ÇнÀÇÒ »ùÇà µ¥ÀÌÅÍ ¼öÀÇ ºÎÁ· 4.2 Ç®°íÀÚ ÇÏ´Â ¹®Á¦¿¡ ºñÇØ º¹ÀâÇÑ ¸ðµ¨À» Àû¿ë 4.3 ÀûÇÕ¼º Æò°¡ ¹× ½ÇÇè ¼³°è(Training, Validation, Test , Cross Validation) 5. Àΰø ½Å°æ¸Á 5.1 ÆÛ¼ÁÆ®·Ð 5.2 ½Å°æ¸Á ¸ðÇüÀÇ ´ÜÁ¡ 6. ¼º´É ÁöÇ¥ Part 03 Deep Learning1. µö·¯´×ÀÇ Á¤ÀÇ 2. µö·¯´×ÀÌ ¹ßÀüÇÏ°Ô µÈ °è±â 3. µö·¯´×ÀÇ Á¾·ù 4. µö·¯´×ÀÇ ¹ßÀüÀ» À̲ö ¾Ë°í¸®Áò 4.1 Dropout 4.2 Activation ÇÔ¼ö 4.3 Batch Normalization 4.4 Initialization 4.5 Optimizer 4.6 AutoEncoder(AE) 4.7 Stacked AutoEncoder 4.8 Denoising AutoEncoder(DAE) Part 04 ÄÄÇ»ÅÍ ºñÀü1. Convolutional Neural Network(CNN) 2. CNN°ú MLP 3. Data Augmentation 4. CNN Architecture 5. Transfer Learning Part 05 ÀÚ¿¬¾î ó¸®1. Data & Task: ¾î¶² µ¥ÀÌÅÍ°¡ ÀÖÀ»±î? 1.1 °¨Á¤ ºÐ¼®(Sentiment Analysis) 1.2 ¿ä¾à(Summarization) 1.3 ±â°è ¹ø¿ª(Machine Translation) 1.4 Áú¹® ÀÀ´ä(Question Answering) 1.5 ±âŸ(etc.) 2. ¹®ÀÚ¸¦ ¼ýÀڷΠǥÇöÇÏ´Â ¹æ¹ý 2.1 Corpus & Out-of-Vocabulary(OOV) 2.2 Byte Pair Encoding(BPE) 2.3 Word Embedding 3. Models 3.1 Deep Learning Models 3.2 Pre-Trained ModelÀÇ ½Ã´ë - Transformer, BERTÀÇ µîÀå 4. Recap 4.1 ?5-3_model_imdb_glove.ipynb¡¯ Äڵ忡 ´ëÇÑ ¼³¸í 4.2 ?5-5_model_imdb_BERT.ipynb¡¯ Äڵ忡 ´ëÇÑ ¼³¸í 4.3 ¸ðµ¨ ¼º´É ºñ±³ Part 06 Other Topics1. Generative Adversarial Networks(GAN) 2. °ÈÇнÀ 3. Domain Adaptation 4. Continual Learning 5. Object Detection 6. Segmentation 7. Meta Learning 8. AutoML