Deep Learning Foundations

Tracks

1. 01🧠Deep Learning 왜?

   0/10 lessons

   Representation, scale, 그리고 안 쓸 때

   Deep learning 은 representation learning 과 differentiable optimization, 충분한 data 또는 pretraining 이 hand-crafted feature 를 이길 때 쓸모 있어.

   Lesson list (10)

   • 01
     Hand-Crafted Feature 의 한계~22 min · features, representation, history
   • 02
     Classical ML vs Deep Learning~22 min · classical-ml, tradeoffs, baselines
   • 03
     Representation Learning~22 min · representation, embeddings, transfer
   • 04
     History Arc~18 min · history, alexnet, transformer
   • 05
     Data, Compute, Hardware~18 min · gpu, tpu, scaling
   • 06
     Neural Network 가 잘하는 것~16 min · use-cases, applications
   • 07
     Deep Learning 이 Overkill 인 곳~16 min · anti-patterns, tradeoffs
   • 08
     Deep Learning 의 비용~16 min · cost, infra, ops
   • 09
     Modern Stack~16 min · pytorch, huggingface, ecosystem
   • 10
     Roadmap: 뭘 배울지~12 min · roadmap, tracks

   ✓퀴즈 · 4 문제→

2. 02🧩Neuron, Layer, Forward Pass

   0/8 lessons

   Linear map 에서 output head 까지

   Network 는 명시적 shape contract 가진 composed transformation.

   Lesson list (8)

   • 01
     Artificial Neuron~22 min · neuron, linear, activation
   • 02
     Perceptron 기초~18 min · perceptron, history
   • 03
     Single Linear Layer 의 한계~18 min · linearity, xor, depth
   • 04
     Multilayer Perceptron~25 min · mlp, fully-connected, depth
   • 05
     Activation Function~22 min · relu, gelu, softmax
   • 06
     Width, Depth, Expressive Power~18 min · capacity, depth, scaling
   • 07
     Forward Pass~22 min · forward, shapes, autograd
   • 08
     Batch, Logit, Output Head~22 min · batches, logits, heads

   ✓퀴즈 · 4 문제→

3. 03⛓️Loss, Gradient, Backpropagation

   0/9 lessons

   Neural network 가 실제 어떻게 학습

   학습은 gradient, optimizer state, stable update rule 통해 loss 줄이기.

   Lesson list (9)

   • 01
     학습한다는 게 뭔가~18 min · learning, loss, gradient-descent
   • 02
     Loss Function~22 min · loss, cross-entropy, mse
   • 03
     Gradient 직관~18 min · gradients, calculus, geometry
   • 04
     Gradient Descent 와 Mini-Batch~22 min · sgd, minibatch, noise
   • 05
     Backpropagation: Chain Rule~25 min · backprop, chain-rule, autograd
   • 06
     Vanishing 과 Exploding Gradient~22 min · vanishing, exploding, depth
   • 07
     Optimizer: SGD 부터 AdamW 까지~22 min · optimizer, adamw, momentum
   • 08
     Learning Rate Schedule~22 min · lr, schedule, warmup
   • 09
     Training Loop~25 min · training-loop, epochs, validation

   ✓퀴즈 · 4 문제→

4. 04🏋️Training Data, State, Checkpoint

   0/8 lessons

   Dataloader, mode, 저장, evaluation

   Training 은 hero 한 코드 cell 이 아니라 stateful workflow.

   Lesson list (8)

   • 01
     Dataset 객체~18 min · dataset, pytorch, data
   • 02
     DataLoader 와 Batching~20 min · dataloader, batching, workers
   • 03
     Device 와 Tensor 배치~18 min · device, cuda, mps
   • 04
     한 Train Step~18 min · train-step, zero-grad, step
   • 05
     Evaluation Mode~16 min · eval, dropout, batchnorm
   • 06
     Training 중 Metric~18 min · metrics, logging, monitoring
   • 07
     Checkpoint~18 min · checkpoint, save, resume
   • 08
     Reproducibility 와 Seed~16 min · seed, reproducibility, determinism

   ✓퀴즈 · 4 문제→

5. 05🛡️Stability, Regularization, Training 작동시키기

   0/10 lessons

   Overfitting, curve, normalization, residual

   Deep learning work 의 대부분이 training 을 stable 하고 정직하게 만드는 거.

   Lesson list (10)

   • 01
     Neural Network 의 Overfitting~18 min · overfitting, generalization, capacity
   • 02
     Loss Curve 읽기~18 min · debugging, loss-curves
   • 03
     Dropout~18 min · dropout, regularization
   • 04
     Weight Decay 와 Early Stopping~18 min · weight-decay, early-stopping, l2
   • 05
     BatchNorm 과 LayerNorm~22 min · batchnorm, layernorm, normalization
   • 06
     Initialization: Xavier 와 He~18 min · init, xavier, he
   • 07
     Residual Connection~22 min · residual, skip, resnet
   • 08
     Gradient Clipping~14 min · clipping, stability
   • 09
     Data Augmentation~18 min · augmentation, regularization
   • 10
     Mixed Precision 과 Reproducibility~18 min · amp, fp16, bf16

   ✓퀴즈 · 4 문제→

6. 06🏗️Architecture: Vision, Sequence, Attention

   0/12 lessons

   CNN, recurrence, attention, transformer

   Architecture 가 data structure 에 대한 가정 encode.

   Lesson list (12)

   • 01
     Image 가 왜 다른가~16 min · images, spatial, translation-invariance
   • 02
     Convolution: Shared Filter~22 min · conv, filters, stride
   • 03
     CNN 직관: Edge 에서 Object 로~18 min · cnn, feature-hierarchy, interpretability
   • 04
     Classic CNN 계보~18 min · cnn, history, resnet
   • 05
     Vision 의 Transfer Learning~18 min · transfer-learning, fine-tuning, vision
   • 06
     Vision Transformer~18 min · vit, transformer, patches
   • 07
     실전 Vision Task~16 min · detection, segmentation, classification
   • 08
     Sequence Data 가 왜 어려운가~16 min · sequence, rnn, context
   • 09
     Recurrent Neural Network~18 min · rnn, lstm, gru
   • 10
     LSTM 과 GRU~18 min · lstm, gru, gating
   • 11
     Attention Breakthrough~22 min · attention, self-attention, transformer
   • 12
     Transformer Block~22 min · transformer, blocks, ffn

   ✓퀴즈 · 4 문제→

7. 07🧪Modern Deep Learning Practice

   0/8 lessons

   PyTorch mental model, transfer, accelerator, tracking

   Practice 가 deep learning 을 notebook stunt 가 아니라 repeatable workflow 로 만드는 곳.

   Lesson list (8)

   • 01
     PyTorch Mental Model~22 min · pytorch, mental-model, modules
   • 02
     Training Loop 해부~18 min · training-loop, scaffold
   • 03
     Transfer Learning vs Fine-Tuning~16 min · transfer, fine-tune, freezing
   • 04
     LoRA: Parameter-Efficient Tuning~22 min · lora, peft, fine-tune
   • 05
     Accelerator 와 Device~16 min · gpu, tpu, mps
   • 06
     Mixed Precision~16 min · amp, bf16, fp16
   • 07
     Experiment Tracking~18 min · tracking, wandb, mlflow
   • 08
     Compilation 과 Export Path~18 min · onnx, torchscript, compile

   ✓퀴즈 · 4 문제→

8. 08🗺️올바른 Approach 고르기

   0/8 lessons

   Scale, foundation model, safety, final map

   Final skill 이 deep learning 이 올바른 도구일 때와 어떤 종류의 deep learning 쓸지 아는 거.

   Lesson list (8)

   • 01
     Deep Learning 이 Classical ML 이길 때~16 min · choice, tradeoffs, classical-ml
   • 02
     MLP, CNN, Transformer, Foundation Model 고르기~18 min · choice, architecture
   • 03
     Parameter Count 와 Scaling~16 min · scaling, parameter-count, compute
   • 04
     Foundation Model~18 min · foundation-model, llm, transfer
   • 05
     Fine-Tuning vs Prompting vs Feature Extraction~16 min · adaptation, tradeoffs
   • 06
     Robustness, Bias, Safety~18 min · safety, bias, robustness
   • 07
     물 안 빠지고 current 유지~16 min · learning, research, habits
   • 08
     Final Map: 기초에서 Modern AI 까지~22 min · recap, roadmap

   ✓퀴즈 · 5 문제→