Present the idea.
Practice the reasoning.
Presentation sheets for teaching in a lecture, and practice sheets for exploring the same ideas through questions, calculations, and interactive visuals.
Lecture presentation sheets
Teach from the page: concise theory, live plots, textbook figures, and discussion prompts.
LSTM, GRU & bidirectional RNNs
Keep the clue. Read the context. Five guided animations, numerical gate experiments, fifteen checks and a design challenge on memory and available context.
Open presentation sheet →Lecture presentation · NewSequence modelling: remember what came before
Sequential data, hidden state, RNN unrolling and long-term dependencies. Six guided animations, numerical experiments, fifteen understanding checks and a printable lecture pack.
Open presentation sheet ↗Lecture presentationTransfer learning
Pretrained models, feature extraction, layer freezing, augmentation, and why vision transformers connect distant evidence. Five visual labs and lecture assets.
Open presentation sheet ↗Lecture presentationCNN architectures: depth, scale & efficiency
Shattered gradients, residual connections, ResNet, Inception, and depthwise separable convolutions. Five interactive explorations with numerical feature maps.
Open presentation sheet ↗Lecture presentationRegularisation
L2, L1, dropout, batch normalisation, and early stopping. Reveal derivations and figures as the lecture unfolds.
Open presentation sheet ↗Lecture presentationWeight initialization
Explore symmetry, signal scale, Xavier, and He with rotatable 3D visuals, guided animations, and eleven checks.
Open presentation sheet ↗ Lecture presentationConvolutional networks
Kernels, stride, padding, feature maps, pooling, receptive field, translation, AlexNet and VGG — typed checks and live grids.
Open presentation sheet ↗Practice sheets
Work through examples during a lecture or revisit them independently. Feedback and progress support learning; these are not graded assessments.