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
Entering the World of Deep Learning
- Defining deep learning and distinguishing it from conventional machine learning approaches
- Practical applications in visual analysis, text processing, and other domains
- A survey of the deep learning landscape: TensorFlow 2.x, Keras, and PyTorch
- Configuring a development workspace with GPU support
Core Mechanics of Deep Learning
- Artificial neurons, activation mechanisms, and network structures
- Forward propagation processes and generating predictions
- Loss metrics used for classification and regression challenges
- Gradient descent optimization and the backpropagation algorithm
- Building your initial neural network using the MNIST dataset
Convolutional Neural Networks for Visual Tasks
- Grasping convolution operations, filters, and feature map generation
- Pooling mechanisms and reducing data dimensions
- CNN frameworks: concepts behind LeNet, VGG, and ResNet
- Designing and training a CNN for image identification
- Analyzing learned features and intermediate layer outputs
Data Augmentation and Enhancing Precision
- The role of data augmentation in preventing overfitting and boosting generalization
- Image manipulation techniques: rotating, flipping, zooming, and cropping
- Creating augmentation workflows using Keras preprocessing layers
- Utilizing dropout, batch normalization, and other regularization methods
- Tracking progress via validation metrics and implementing early stopping strategies
Leveraging Transfer Learning with Existing Models
- Conceptualizing transfer learning and its effectiveness
- Importing pre-trained models from Keras Applications (such as ResNet, EfficientNet, MobileNet)
- Feature extraction: locking base layers while training new classification heads
- Fine-tuning: selectively unlocking layers to adapt to specific domains
- Drawing high accuracy even when training data is limited
Recurrent Networks and Handling Sequences
- Introduction to sequential information and time-based relationships
- Recurrent Neural Networks (RNNs) and addressing the vanishing gradient issue
- Employing LSTM and GRU units for long-term dependencies
- Training a model capable of generating character-level text
- Word embeddings and utilizing the Embedding layer within Keras
Foundations of Natural Language Processing
- Text preparation: tokenization, padding sequences, and constructing vocabularies
- Creating a text classification model using RNNs and LSTMs
- Concepts behind sequence-to-sequence models for translation tasks
- The significance of attention mechanisms in contemporary NLP
- Practical NLP implementation using TensorFlow 2.x text utilities
Capstone Project: Image Caption Generation
\r- Merging visual processing and language understanding in a hybrid architecture
- Gathering image features via a pre-trained CNN encoder
- Developing an LSTM-based decoder for writing captions
- Handling multiple input streams using the Keras functional API
- Executing training and assessing the complete captioning workflow
Moving Forward and Additional Resources
- Deploying finished models using TensorFlow Serving
- Investigating transformer designs and large-scale language models
- NVIDIA DLI advanced workshops and certification routes
- Community hubs, public datasets, and project inspiration
Requirements
- Elementary competence in Python coding (understanding functions, loops, dictionaries, and arrays)
- Knowledge of core programming principles including variables, conditional logic, and data structures
- No previous background in deep learning or machine learning is necessary
Target Audience
- Software engineers moving into the fields of AI and machine learning
- Data analysts and scientists looking to acquire deep learning expertise
- Technical staff interested in comprehending and implementing neural network frameworks
- Academic learners and researchers embarking on their deep learning education
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
