A hands-on collection of deep learning labs covering core concepts — from building neural networks from scratch to transformers and vision models. Each lab is designed to reinforce theoretical understanding through practical implementation.
dl-labs/
├── lab01/ # Feed Forward & Back-Propagation (from scratch)
├── lab02/ # ANN for MNIST Classification
├── lab03/ # CNN for MNIST + Comparative Report
├── lab04/ # ResNet-34 for Skin Lesion & Deepfake Detection
├── lab05/ # AutoEncoders — MNIST Compression
├── lab06/ # Anomaly Detection using VAE & GAN (MRI)
├── lab07/ # Tumor Segmentation using UNet
├── lab08/ # Sentiment Classification using RNN & LSTM
├── lab09/ # News Summarization using BART (Transformers)
└── lab10/ # Chest X-Ray Classification using ViT
| Lab | Topic | Dataset |
|---|---|---|
| 01 | Feedforward & Backpropagation from scratch | IRIS (Setosa vs Versicolor) |
| 02 | Fully Connected ANN | MNIST |
| 03 | CNN + ANN vs CNN comparison report | MNIST |
| 04 | ResNet-34 (custom) — Skin lesion & Deepfake detection | ISIC 2019, Custom Deepfake |
| 05 | PCA vs AutoEncoder compression, t-SNE visualization | MNIST |
| 06 | Anomaly detection — VAE & GAN reconstruction | LGG MRI |
| 07 | UNet segmentation — Baseline vs Heatmap-guided | LGG MRI |
| 08 | Sentiment analysis — RNN vs LSTM | SST-2 |
| 09 | Fine-tuned BART for news summarization | ILSUM-1.0 (English) |
| 10 | Vision Transformer (ViT) for chest X-ray classification | Chest X-Ray Dataset |
pip install torch torchvision
pip install tensorflow keras
pip install transformers datasets
pip install scikit-learn matplotlib numpy pandasMost labs are designed to run on Google Colab / Kaggle free-tier GPUs.
IIITDM Kancheepuram — Deep Learning Course