Course Objective & Scope
Defining a neural network takes only a few lines of library code, yet understanding why gradients vanish, why the loss oscillates or why a network memorises training data requires a solid grounding. This programme covers multilayer perceptrons (MLPs), ReLU/GELU activations, cross-entropy and MSE losses, backpropagation with autograd, SGD/AdamW optimisation and regularisation via dropout, weight decay and batch/layer normalisation.
Starting from tensor operations in PyTorch, participants build DataLoaders, training/validation loops and introductory CNN/sequence blocks from scratch, producing a reusable working example that benchmarks a neural model against a classical ML baseline.
Who Should Attend?
- Software developers and data scientists moving into deep learning
- Analytics specialists stepping from classical ML to neural networks
- R&D and platform engineers joining AI project teams
- Prerequisite: Python programming and basic linear algebra/calculus
Key Takeaways
- Work fluently with tensors and automatic differentiation (autograd) graphs
- Select network layers, activations and loss functions suited to the task
- Write reproducible training and validation loops in PyTorch
- Control overfitting using learning rate schedules, dropout and normalisation
- Diagnose common training failures by inspecting loss curves and gradient flow
Curriculum
Neural Networks and the Autograd Mechanism
- Tensors, linear layers and non-linear activations (ReLU, GELU, SiLU)
- Computation graphs, the chain rule and backpropagation
Loss Functions and Optimisation
- MSE, Huber, Binary and Categorical Cross-Entropy
- SGD, momentum, Adam and AdamW; learning-rate behaviour
PyTorch Training Loops and Data Pipelines
- Dataset, DataLoader, mini-batching and GPU placement
- Train/eval modes, checkpointing and deterministic seeding
Regularisation and Training Stability
- Diagnosing overfitting, weight decay, dropout and early stopping
- Weight initialisation (He/Xavier), BatchNorm and LayerNorm
Intro to Spatial and Sequential Layers & Case Lab
- Intuition behind convolutional (CNN) and Embedding layers
- End-to-end case implementation and training checklist
Other Courses in This Category
-
Stable, large-scale deep learning covering BF16/FP16 mixed precision, multi-GPU training with DDP/FSDP, self-supervised and contrastive pretraining, metric learning, knowledge distillation and GPU profiling.
- Live Course
- 5 modules
-
From ResNet and ConvNeXt to Vision Transformers (ViT), Swin, hybrid backbones, State Space Models (Mamba), Mixture-of-Experts (MoE) and multimodal fusion: selecting and designing architectures for your problem and hardware.
- Live Course
- 5 modules
-
Self-attention mechanics, Multi-Head and Grouped-Query Attention, RoPE positional encodings, encoder/decoder/encoder-decoder families, KV caching and FlashAttention, implemented from scratch in PyTorch and with Hugging Face.
- Live Course
- 5 modules
-
The Keras 3 and TensorFlow 2 ecosystem: Functional/Subclassing APIs, high-throughput tf.data pipelines, custom GradientTape loops, tf.distribute and enterprise deployment via SavedModel, TF Serving and TFLite.
- Live Course
- 5 modules
-
Quantisation (PTQ, QAT, INT8/INT4), structured pruning, knowledge distillation and deployment with ONNX Runtime, TensorRT, OpenVINO and LiteRT (TFLite) for lower cloud GPU costs and fast edge inference.
- Live Course
- 5 modules
-
Foundations and architectures of deep generative modelling: VAEs, GANs, Normalizing Flows, Denoising Diffusion (DDPM/Latent Diffusion), Flow Matching and autoregressive models for synthetic data and multimodal generation.
- Live Course
- 5 modules
Get in touch
