\n \ub525\ub7ec\ub2dd(Deep Learning)\uc740 \uc778\uacf5\uc9c0\ub2a5(Artificial Intelligence)\uacfc \uba38\uc2e0\ub7ec\ub2dd(Machine Learning)\uc758 \ud55c \ubd84\uc57c\ub85c,
\n \ub370\uc774\ud130\uc5d0\uc11c \ud328\ud134\uc744 \ud559\uc2b5\ud558\uc5ec \uc608\uce21 \ubaa8\ub378\uc744 \ub9cc\ub4dc\ub294 \uae30\uc220\uc785\ub2c8\ub2e4. \ucd5c\uadfc \uba87 \ub144 \uc0ac\uc774\uc5d0 \ube45\ub370\uc774\ud130\uc640
\n \ucef4\ud4e8\ud305 \ud30c\uc6cc\uc758 \ubc1c\uc804\uc73c\ub85c \ub9ce\uc740 \ubd84\uc57c\uc5d0\uc11c \uc8fc\ubaa9\ubc1b\uace0 \uc788\uc73c\uba70, \ud2b9\ud788 \ucef4\ud4e8\ud130 \ube44\uc804, \uc790\uc5f0\uc5b4 \ucc98\ub9ac, \uc74c\uc131 \uc778\uc2dd \ub4f1
\n \ub2e4\uc591\ud55c \ubd84\uc57c\uc5d0 \ud65c\uc6a9\ub418\uace0 \uc788\uc2b5\ub2c8\ub2e4. \ud558\uc9c0\ub9cc \ub525\ub7ec\ub2dd \ubaa8\ub378\uc740 \uc124\uacc4 \ubc0f \ud559\uc2b5 \uacfc\uc815\uc5d0\uc11c \uc5ec\ub7ec \uac00\uc9c0 \ubb38\uc81c\uc810\ub4e4\uc774 \ubc1c\uc0dd\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.
\n \ubcf8 \ubb38\uc11c\uc5d0\uc11c\ub294 \ub525\ub7ec\ub2dd\uc758 \uc8fc\uc694 \ubb38\uc81c\uc810\uacfc \uc774\ub97c \ud574\uacb0\ud558\uae30 \uc704\ud55c \ubc29\uc548, \uadf8\ub9ac\uace0 \ud30c\uc774\ud1a0\uce58(PyTorch)\ub97c \ud65c\uc6a9\ud55c \uc608\uc81c \ucf54\ub4dc\uc5d0 \ub300\ud574 \uc54c\uc544\ubcf4\uaca0\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n \uacfc\uc801\ud569\uc740 \ubaa8\ub378\uc774 \ud6c8\ub828 \ub370\uc774\ud130\uc5d0 \ub108\ubb34 \uc798 \uc801\ud569\ub418\uc5b4 \uc0c8\ub85c\uc6b4 \ub370\uc774\ud130\uc5d0 \ub300\ud55c \uc77c\ubc18\ud654 \uc131\ub2a5\uc774 \ub5a8\uc5b4\uc9c0\ub294 \ud604\uc0c1\uc744 \ub9d0\ud569\ub2c8\ub2e4.
\n \uc774\ub294 \uc8fc\ub85c \ub370\uc774\ud130\uac00 \ubd80\uc871\ud558\uac70\ub098 \ubaa8\ub378\uc774 \ub108\ubb34 \ubcf5\uc7a1\ud560 \ub54c \ubc1c\uc0dd\ud569\ub2c8\ub2e4.\n <\/p>\n
\n \ubd84\ub958 \ubb38\uc81c\uc5d0\uc11c \uac01 \ud074\ub798\uc2a4\uc758 \ub370\uc774\ud130 \uc218\uac00 \ubd88\uade0\ud615\ud55c \uacbd\uc6b0, \ubaa8\ub378\uc740 \ub9ce\uc740 \ub370\uc774\ud130\uac00 \uc788\ub294 \ud074\ub798\uc2a4\uc5d0\ub9cc \uc798 \uc801\ud569\ub420 \uc218 \uc788\uc2b5\ub2c8\ub2e4.
\n \uc774\ub85c \uc778\ud574 \uc801\uc740 \ub370\uc774\ud130 \ud074\ub798\uc2a4\uc5d0 \ub300\ud55c \uc131\ub2a5\uc774 \ub5a8\uc5b4\uc9c8 \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n \uc801\uc808\ud55c \ud559\uc2b5\ub960\uc744 \uc120\ud0dd\ud558\ub294 \uac83\uc740 \ubaa8\ub378 \ud559\uc2b5\uc5d0 \ub9e4\uc6b0 \uc911\uc694\ud569\ub2c8\ub2e4. \ud559\uc2b5\ub960\uc774 \ub108\ubb34 \ub192\uc73c\uba74 \uc190\uc2e4 \ud568\uc218\uac00 \ubc1c\uc0b0\ud560 \uc218 \uc788\uace0,
\n \ub108\ubb34 \ub0ae\uc73c\uba74 \uc218\ub834 \uc18d\ub3c4\uac00 \ub290\ub824\uc838 \ud559\uc2b5\uc774 \ube44\ud6a8\uc728\uc801\uc77c \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n \ub525\ub7ec\ub2dd \ubaa8\ub378\uc740 \ube14\ub799\ubc15\uc2a4 \ubaa8\ub378\ub85c, \uadf8 \ub0b4\ubd80 \ub3d9\uc791\uc774\ub098 \uc608\uce21 \uacb0\uacfc\uc5d0 \ub300\ud55c \ud574\uc11d\uc774 \uc5b4\ub824\uc6cc \uae30\uc5c5\uc774\ub098 \uc758\ub8cc \ubd84\uc57c \ub4f1\uc5d0\uc11c
\n \uc2e0\ub8b0\uc131 \ubb38\uc81c\ub97c \uc77c\uc73c\ud0ac \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n \ub300\uaddc\ubaa8 \ubaa8\ub378\uc744 \ud559\uc2b5\ud558\ub824\uba74 \ub9ce\uc740 \uacc4\uc0b0 \uc790\uc6d0\uacfc \uba54\ubaa8\ub9ac\uac00 \ud544\uc694\ud569\ub2c8\ub2e4. \uc774\ub294 \uacbd\uc81c\uc801 \ube44\uc6a9\uacfc \uc5d0\ub108\uc9c0
\n \uc18c\ubaa8\uc758 \ubb38\uc81c\ub97c \uc57c\uae30\ud569\ub2c8\ub2e4.\n <\/p>\n
\n \uacfc\uc801\ud569\uc744 \ubc29\uc9c0\ud558\uae30 \uc704\ud574 \ub2e4\uc591\ud55c \ubc29\ubc95\ub4e4\uc774 \uc0ac\uc6a9\ub429\ub2c8\ub2e4. \uadf8 \uc911 \uc77c\ubd80\ub294 \ub2e4\uc74c\uacfc \uac19\uc2b5\ub2c8\ub2e4:\n <\/p>\n
\n \ub370\uc774\ud130\uc758 \ubd88\uade0\ud615 \ubb38\uc81c\ub97c \ud574\uacb0\ud558\uae30 \uc704\ud55c \ubc29\ubc95\uc73c\ub85c\ub294 \ub2e4\uc74c\uacfc \uac19\uc740 \uae30\ubc95\ub4e4\uc774 \uc788\uc2b5\ub2c8\ub2e4:\n <\/p>\n
\n \ud559\uc2b5 \uc18d\ub3c4\ub97c \ub192\uc774\uae30 \uc704\ud574 \uc801\uc751\ud615 \ud559\uc2b5\ub960 \uc54c\uace0\ub9ac\uc998 (Adam, RMSProp \ub4f1)\uc744 \uc0ac\uc6a9\ud560 \uc218 \uc788\uc73c\uba70,
\n \ubc30\uce58 \uc815\uaddc\ud654 (Batch Normalization)\ub97c \ud65c\uc6a9\ud558\uc5ec \ud559\uc2b5\uc744 \uc548\uc815\ud654\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n \ubaa8\ub378\uc758 \ud574\uc11d \uac00\ub2a5\uc131\uc744 \ub192\uc774\uae30 \uc704\ud55c \ubc29\ubc95\uc5d0\ub294 LIME, SHAP \ub4f1\uacfc \uac19\uc740 \uae30\ubc95\uc744 \uc774\uc6a9\ud558\uc5ec \ubaa8\ub378\uc758 \uc608\uce21
\n \uacb0\uacfc\uc5d0 \ub300\ud55c \ud574\uc11d\uc744 \uc81c\uacf5\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n \ubaa8\ub378 \uc555\ucd95 (Model Compression)\uc774\ub098 \uacbd\ub7c9\ud654 \ub124\ud2b8\uc6cc\ud06c (MobileNet, SqueezeNet \ub4f1)\ub97c \uc0ac\uc6a9\ud558\uc5ec
\n \ubaa8\ub378\uc758 \uaddc\ubaa8\ub97c \uc904\uc774\uace0 \uc2e4\ud589 \uc2dc\uac04\uc744 \ub2e8\ucd95\ud560 \uc218 \uc788\uc2b5\ub2c8\ub2e4.\n <\/p>\n
\n \ub2e4\uc74c\uc740 \ud30c\uc774\ud1a0\uce58\ub97c \uc0ac\uc6a9\ud558\uc5ec \uac04\ub2e8\ud55c \uc2e0\uacbd\ub9dd\uc744 \uad6c\ucd95\ud558\uace0 \ud559\uc2b5\uc2dc\ud0a4\ub294 \uc608\uc81c\uc785\ub2c8\ub2e4.
\n \uc774 \uc608\uc81c\ub294 MNIST \ub370\uc774\ud130\uc14b\uc744 \uc0ac\uc6a9\ud558\uc5ec \uc190\uae00\uc528 \uc22b\uc790\ub97c \ubd84\ub958\ud558\ub294 \ubaa8\ub378\uc744 \uad6c\ud604\ud569\ub2c8\ub2e4.\n <\/p>\n
\nimport torch\nimport torch.nn as nn\nimport torch.optim as optim\nimport torchvision.transforms as transforms\nfrom torchvision import datasets\nfrom torch.utils.data import DataLoader\n <\/code><\/pre>\n3.2. \ud558\uc774\ud37c\ud30c\ub77c\ubbf8\ud130 \uc124\uc815<\/h3>\n\n# \ud558\uc774\ud37c\ud30c\ub77c\ubbf8\ud130 \uc124\uc815\nbatch_size = 64\nlearning_rate = 0.001\nnum_epochs = 5\n <\/code><\/pre>\n3.3. \ub370\uc774\ud130 \uc900\ube44<\/h3>\n\n# \ub370\uc774\ud130\uc14b \uc900\ube44\ntransform = transforms.Compose([transforms.ToTensor(), transforms.Normalize((0.5,), (0.5,))])\ntrain_dataset = datasets.MNIST(root='.\/data', train=True, download=True, transform=transform)\ntest_dataset = datasets.MNIST(root='.\/data', train=False, download=True, transform=transform)\ntrain_loader = DataLoader(dataset=train_dataset, batch_size=batch_size, shuffle=True)\ntest_loader = DataLoader(dataset=test_dataset, batch_size=batch_size, shuffle=False)\n <\/code><\/pre>\n3.4. \ubaa8\ub378 \uc815\uc758<\/h3>\n\n# \uc2e0\uacbd\ub9dd \ubaa8\ub378 \uc815\uc758\nclass SimpleNN(nn.Module):\n def __init__(self):\n super(SimpleNN, self).__init__()\n self.fc1 = nn.Linear(28 * 28, 128) # \uc785\ub825\uce35\n self.fc2 = nn.Linear(128, 64) # \uc740\ub2c9\uce35\n self.fc3 = nn.Linear(64, 10) # \ucd9c\ub825\uce35\n\n def forward(self, x):\n x = x.view(-1, 28 * 28) # Flatten\n x = torch.relu(self.fc1(x))\n x = torch.relu(self.fc2(x))\n x = self.fc3(x)\n return x\n\nmodel = SimpleNN()\n <\/code><\/pre>\n3.5. \uc190\uc2e4 \ud568\uc218\uc640 \uc635\ud2f0\ub9c8\uc774\uc800 \uc124\uc815<\/h3>\n\n# \uc190\uc2e4 \ud568\uc218 \ubc0f \uc635\ud2f0\ub9c8\uc774\uc800 \uc124\uc815\ncriterion = nn.CrossEntropyLoss()\noptimizer = optim.Adam(model.parameters(), lr=learning_rate)\n <\/code><\/pre>\n3.6. \ubaa8\ub378 \ud559\uc2b5<\/h3>\n\n# \ubaa8\ub378 \ud559\uc2b5\nfor epoch in range(num_epochs):\n for images, labels in train_loader:\n optimizer.zero_grad() # \uae30\uc6b8\uae30 \ucd08\uae30\ud654\n outputs = model(images) # \uc608\uce21\n loss = criterion(outputs, labels) # \uc190\uc2e4 \uacc4\uc0b0\n loss.backward() # \uc5ed\uc804\ud30c\n optimizer.step() # \uac00\uc911\uce58 \uc5c5\ub370\uc774\ud2b8\n print(f'Epoch [{epoch+1}\/{num_epochs}], Loss: {loss.item():.4f}')\n <\/code><\/pre>\n3.7. \ubaa8\ub378 \ud3c9\uac00<\/h3>\n\n# \ubaa8\ub378 \ud3c9\uac00\nmodel.eval() # \ud3c9\uac00\ubaa8\ub4dc\ub85c \uc804\ud658\ncorrect = 0\ntotal = 0\nwith torch.no_grad():\n for images, labels in test_loader:\n outputs = model(images)\n _, predicted = torch.max(outputs.data, 1)\n total += labels.size(0)\n correct += (predicted == labels).sum().item()\n\nprint(f'Accuracy of the model on the test images: {100 * correct \/ total:.2f}%')\n <\/code><\/pre>\n3.8. \uacb0\ub860<\/h3>\n
\n \ubcf8 \uac15\uc88c\uc5d0\uc11c\ub294 \ub525\ub7ec\ub2dd\uc758 \uc5ec\ub7ec \ubb38\uc81c\uc810\uacfc \ud574\uacb0 \ubc29\uc548\uc5d0 \ub300\ud574 \ub17c\uc758\ud558\uace0, \ud30c\uc774\ud1a0\uce58\ub85c \uac04\ub2e8\ud55c
\n \uc2e0\uacbd\ub9dd \ubaa8\ub378\uc744 \uad6c\ud604\ud574 \ubcf4\uc558\uc2b5\ub2c8\ub2e4. \ub525\ub7ec\ub2dd \ubaa8\ub378\uc744 \uc131\uacf5\uc801\uc73c\ub85c \uc6b4\uc601\ud558\uae30 \uc704\ud574\uc11c\ub294
\n \ubb38\uc81c\uc758 \ud2b9\uc131\uc744 \ud30c\uc545\ud558\uace0, \ub2e4\uc591\ud55c \uae30\ubc95\uc744 \uc801\uc808\ud788 \uc870\ud569\ud558\uc5ec \ucd5c\uc801\uc758 \ubaa8\ub378\uc744 \ub3c4\ucd9c\ud558\ub294
\n \uacfc\uc815\uc774 \ud544\uc694\ud569\ub2c8\ub2e4.\n <\/p>\n
\n \uc55e\uc73c\ub85c \ub525\ub7ec\ub2dd \uae30\uc220\uc774 \ub354\uc6b1 \ubc1c\uc804\ud568\uc5d0 \ub530\ub77c, \uc6b0\ub9ac\uc758 \uc0b6\uc5d0 \ub354\uc6b1 \uae4a\uc219\uc774 \ub4e4\uc5b4\uc62c \uac83\uc73c\ub85c \uae30\ub300\ub429\ub2c8\ub2e4.
\n \uc774\ub97c \uc704\ud574 \uc9c0\uc18d\uc801\uc778 \uc5f0\uad6c\uc640 \uc801\uc6a9\uc774 \ud544\uc694\ud558\uba70, \uc774\ub7ec\ud55c \uacfc\uc815\uc5d0\uc11c \ub9ce\uc740 \uac1c\ubc1c\uc790\ub4e4\uc774 \ub2e4\uc591\ud55c
\n \ubb38\uc81c\ub4e4\uc744 \ud574\uacb0\ud574 \ub098\uac00\uae30\ub97c \ud76c\ub9dd\ud569\ub2c8\ub2e4.\n <\/p>\n
<\/body><\/p>\n","protected":false},"excerpt":{"rendered":"