Deep Learning | CS PPT +Worksheets +Activity + Mcq Quiz + Answers_Marking Scheme

PPT – Core Lecture (13 slides)
Dark neural-network-themed design with live ANN diagrams drawn in-slide. Covers: ANN structure with interactive neuron/synapse visuals → Feature hierarchy (pixels→edges→shapes→objects) → Reinforcement Learning loop diagram (Agent ↔ Environment) → RL & DL use cases → All 6 models (CNN, RNN, LSTM, GAN, Transformer, Diffusion) → 6 Frameworks (TensorFlow, PyTorch, Keras, HuggingFace, ONNX, CUDA) → Encoder-Decoder with attention diagram → 4 Real-time worked examples → Limitations vs Future advancements grid → Practice + Summary.

Worksheet (61 marks)
Five sections: ANN fundamentals (loss, backprop, activation), deep learning models fill-in table, Reinforcement Learning (including RLHF), encoder-decoder + frameworks, and evaluation questions on limitations and future. Full ruled answer lines throughout.

Activity – AI Design Studio Challenge (~50 min)
Groups act as an AI startup with one of four briefs (MediScan AI / AutoDrive UK / SoundForge AI / TutorBot). They select models and frameworks, sketch a system diagram, complete a structured risk assessment table (risk → impact → mitigation), and pitch to the class with peer review checklist. Extension: research Explainable AI (XAI).

MCQ Quiz (12 questions)
Covers what makes DL “deep”, activation functions, model selection, RL rewards, Transformer architecture, encoder role, PyTorch vs TensorFlow, LSTM advantages, GAN structure, black-box problem, RLHF, and federated learning.

Answers & Marking Scheme
Full model answers with per-mark breakdowns for all 61 worksheet marks, complete MCQ key with technical explanations, and detailed activity marking guidance including how to assess model justification and risk tables.