Thank you for sending your enquiry! One of our team members will contact you shortly.
Thank you for sending your booking! One of our team members will contact you shortly.
Course Outline
Introduction to Deep Learning
- What is deep learning and how it differs from traditional machine learning.
- Real-world applications in computer vision, NLP, and beyond.
- Overview of the deep learning ecosystem: TensorFlow 2.x, Keras, PyTorch.
- Setting up a GPU-accelerated development environment.
The Mechanics of Deep Learning
- Artificial neurons, activation functions, and network layers.
- Forward propagation and computing predictions.
- Loss functions for classification and regression tasks.
- Gradient descent optimization and backpropagation.
- Training your first neural network on the MNIST dataset.
Convolutional Neural Networks for Computer Vision
- Understanding convolution, filters, and feature maps.
- Pooling layers and dimensionality reduction.
- CNN architectures: LeNet, VGG, and ResNet concepts.
- Building and training a CNN for image classification.
- Visualizing learned features and intermediate activations.
Data Augmentation and Improving Model Accuracy
- Why data augmentation combats overfitting and improves generalization.
- Image transformations: rotation, flipping, zooming, and cropping.
- Implementing augmentation pipelines with Keras preprocessing layers.
- Dropout, batch normalization, and other regularization techniques.
- Monitoring training with validation metrics and early stopping.
Transfer Learning with Pre-Trained Models
- Understanding transfer learning and why it works.
- Loading pre-trained models from Keras Applications (ResNet, EfficientNet, MobileNet).
- Feature extraction: freezing base layers and training new classifiers.
- Fine-tuning: selectively unfreezing layers for domain adaptation.
- Achieving high accuracy with limited training data.
Recurrent Networks and Sequence Modeling
- Introduction to sequential data and temporal dependencies.
- Recurrent neural networks (RNNs) and the vanishing gradient problem.
- LSTM and GRU cells for long-range dependencies.
- Training a character-level text generation model.
- Word embeddings and the Embedding layer in Keras.
Natural Language Processing Fundamentals
- Text preprocessing: tokenization, padding, and vocabulary building.
- Building a text classifier with RNNs and LSTMs.
- Sequence-to-sequence models for machine translation concepts.
- Attention mechanisms and their role in modern NLP.
- Practical NLP with TensorFlow 2.x text processing APIs.
Final Project: Image Captioning
- Combining computer vision and NLP in a multimodal architecture.
- Extracting image features with a pre-trained CNN encoder.
- Building an LSTM-based decoder for caption generation.
- Managing multiple input layers in Keras functional API.
- Training and evaluating the end-to-end captioning pipeline.
Next Steps and Resources
- Deploying trained models with TensorFlow Serving.
- Exploring transformer architectures and large language models.
- NVIDIA DLI advanced workshops and certification pathways.
- Community resources, datasets, and project ideas.
Requirements
- Basic proficiency in Python programming (functions, loops, dictionaries, arrays).
- Familiarity with programming concepts such as variables, conditionals, and data structures.
- No prior experience in deep learning or machine learning is required.
Target Audience
- Software developers and engineers transitioning into AI and machine learning.
- Data analysts and data scientists looking to acquire deep learning skills.
- Technical professionals seeking to understand and apply neural network models.
- Students and researchers starting their journey in deep learning.
8 Hours
Testimonials (2)
The training was organized and well-planned out, and I come out of it with systematized knowledge and a good look at topics we looked at
Magdalena - Samsung Electronics Polska Sp. z o.o.
Course - Deep Learning with TensorFlow 2
I really liked the end where we took the time to play around with CHAT GPT. The room was not set up the best for this- instead of one large table a couple of small ones so we could get into small groups and brainstorm would have helped
