Stat 654: Introduction to Applied Deep Learning
Department of Statistics and Data Science, CSU East Bay
Spring 2026:
Canvas problem: Resolved
Canvas problem: The university is having problems with Canvas. Hopefully the problems will be solved soon. If you are trying to submit work in this class (or any other class for me) please do so when Canvas is accessible again. DO NOT email your work.
If the problem continues beyond Wednesday next week I will email with instructions on how to submit your work.
Week 8: Finals Week
- Finals week. I will have office hours online on Tuesday and Thursday.
- Monday: The scheduled time for the Final Exam for this class is on Monday at 12:45. I will log into my Zoom Office Hour link at 12:45 for final questions. If no one logs in during the first 30 minutes I will leave the Zoom meeting and you can email me with any questions.
- Spotlight course: UNC bios740
- Spotlight course: Andrej Karpathy Zero To Hero Youtube Playlist
- Spotlight website: Cerabras MOE
- Spotlight LLM: DeepSeek
- Spotlight software: Ultralytics
Week 7:
- Final: The final for the class will be take-home. You can complete it at home individually. I will be available during the scheduled exam time for questions. See the Homework link.
- Project: Your project should be completed before the end of the semester. The last day you can submit is May 15.
- Spotlight website: Using Pre-Trained Models
- Spotlight website: Posit Keras Using Pre-Trained Models
- Spotlight website: Tensorflow Hub tfhub
- Code Example:
- R notebook: 11_Yolo.zip
- Spotlight Transformers: Hugging Face Hugging Face Learn NLP Course Deep RL Course
- Spotlight LLMs: Cohere Cohere LLM University
- RStudio Keras website: keras3 cheatsheet
- Spotlight blog: Posit AI Blog
- Spotlight gallery: Posit AI Gallery
- Spotlight wiki: pathmind.ai pathmind.ai glossary
- Spotlight blog: Top 10 Pretrained Models to get you Started with Deep Learning
- Spotlight blog: Using Keras’ Pre-trained Models for Feature Extraction in Image Clustering
- Spotlight blog: Common architectures in convolutional neural networks. What is ResNet?
- What are Autoencoders? ArXiv Insights Variational Autoecoders
- What are GANs? ArXiv Insights GANS
- What is Reinforcement Learning? ArXiv Insights Reinforcement Learning keras-rl
- What is Stable Diffusion? Latent Diffusion Models EXPLAINED
- What is AutoKeras? autokeras
- What are LLMs? What is a Large Language Model (LLM)?
- What is autoGPT? What is Auto-GPT? Everything to know about the next powerful AI tool
- Presentation:
- Spotlight blog: NVIDIA Upscaling
- Spotlight software: NVIDIA RTX
- Spotlight translation hardware: iflytek
- Spotlight paper: LSTM Neural Reordering Feature for Statistical Machine Translation
- Spotlight paper: Translation with Attention
- Spotlight software: magenta MusicVAE MusicML Jukebox MuseNet Riffusion
- Spotlight conference: Google I/O
- Warning: “Godfather of artificial intelligence” leaves Google to talk about the tech’s potential dangers
- Warning: Titans of AI Andrew Ng and Yann LeCun oppose call for pause on powerful AI systems
- Spotlight book: Deep Learning This is a bit more advanced book.
- Spotlight talk: Deep Learning
- Spotlight Website: deeplearning.ai
- Spotlight LLM: groq.com
- Spotlight LLM: cohere Comand-R+ LLMU
- Spotlight LLM: OpenAI
- Spotlight AI: Ollama
- Spotlight AI: Unsloth Studio
Week 6:
- Warning: There is a new version of R available. Be advised not to update your R before the end of the semester. It would be better to update your install of R after class is over to avoid any problems with your installed packages we are using in the class.
- Spotlight software: RStudio has released a new version of the RStudio IDE that includes Posit Assistant. It gives a 1 month free trial.
- Spotlight software: Rtools
- Homework: Homework 5 has been posted. And Homework 6 has been posted, but will not be collected.
- Project: The class Project has been assigned. Make a stylized picture.
- Spotlight blog: Visualizing Neural Networks with the Grand Tour
- YouTube: vcubingx
- Spotlight blog: Understanding LSTM Networks
- YouTube: 3Blue1Brown Neural Networks
- arXiv paper: An Introduction to Transformers
- arXiv paper: Attention Is All You Need
- Spotlight Transformers: Hugging Face
- Spotlight book: Natural Language Processing with Transformers, Revised Edition
- Spotlight book: Transformers for Natural Language Processing - Second Edition
- Spotlight software: PyTorch
- Spotlight Course: Google Machine Learning Text Classification
- Spotlight Blog: Neural Network Embeddings Explained
- Spotlight: BERT
- Spotlight paper: GloVe
- Code Examples: Chapter 11
- Case Studies: Chapter 12
- LSTM to write sentences
- Transformers
- Deep Dream Elephants
- Neural style transfer
- Generating Images
- GANs
- Google Tensorflow: Tensorflow Resources
- Spotlight blog: How to Use t-SNE Effectively
- Spotlight Conference: AI Accelerator Summit
- Spotlight Github: Neural-Style-Transfer download one picture and one style. try on CoLab.
- Example Style Transfer:
Week 5:
- Quiz: Code Suggestions using only x1 and x2, for your solution you should use some or all of the other variables.
- lastname_firstname_Stat654_quiz_keras_ver01a.qmd
- lastname_firstname_Stat654_quiz_keras_ver01a.html
- lastname_firstname_Stat654_quiz_keras_ver01b.qmd
- lastname_firstname_Stat654_quiz_keras_ver01b.html
- lastname_firstname_Stat654_quiz_keras_ver02a.qmd
- lastname_firstname_Stat654_quiz_keras_ver02a.html
- lastname_firstname_Stat654_quiz_keras_ver02b.qmd
- lastname_firstname_Stat654_quiz_keras_ver02b.html
- lastname_firstname_Stat654_quiz_torch.qmd torch backend from tidymodels
- lastname_firstname_Stat654_quiz_torch.html
- lastname_firstname_Stat654_quiz_keras_ver01a_python.qmd
- lastname_firstname_Stat654_quiz_keras_ver01a_python.html
- lastname_firstname_Stat654_quiz_keras_ver01a.ipynb Colab Notebook
- Midterm: The take-home Midterm will be given this week. See the Assignments link.
- Project: The Project will be given this week.
- Code Examples:
- dogs-vs-cats.zip This is a very large file. This file should probably be downloaded to your computer outside of class and on a wired connection.
- Warning: Starting with Chapter 8 the code becomes very very computationally intensive. Unless you have a very recent computer with lots of CPU processor cores or an NVidia graphics card and tensorflow[gpu] set up you should change the number of epochs to 5 to see how long each part of the code takes to run before trying this code.
- Presentation:
- Spotlight website: ImageNet
- Presentation:
Week 4:
- Quiz: The take-home Quiz has been posted. See the Assignments link.
- Homework: Homework 4 has been posted.
- Presentation:
- Code Example:
- Spotlight website:
- Spotlight website: kaggle dogs-vs-cats
- Code Examples:
- dogs-vs-cats_ver05.R 3rd Edition
- cats_vs_dogs_dir_ver04.R 2nd Edition
- cats_vs_dogs_dir_ver03.R 1st Edition
- dogs-vs-cats.zip This is a very large file. This file should probably be downloaded to your computer outside of class and on a wired connection.
- Spotlight YouTube: Convolutional Neural Networks Explained (CNN Visualized)
- Spotlight blog:
- Spotlight Software:
- Code Example:
- Run LLMs Locally:
- Links:
Week 3:
- Homework: Homework 3 has been posted.
- Google Learning:
- Code Example: We will look at the R code from from Chapter 4.
- Binary classification, movie_sentiment.qmd
- Basic Text Classification
- Text classification from scratch
- Multi-class classification, newswire_classification.qmd
- Basic Text Classification
- Regression, boston_housing.qmd
- Basic Regression
- Basic Image Classsification Fashion MNIST fashion01.R fashion01.R - Colab
- Code Example: We will look at the R code from from Chapter 4. Colab Notebooks Being Reviewed keras3
- Binary classification, classifying-movie-reviews.R - CoLab
- Multi-class classification, classifying-newswires.R CoLab
- Regression, predicting-house-prices.R - CoLab
- Presentation:
- Spotlight book: Deep Learning This is a bit more advanced book.
- Spotlight talk: Deep Learning
- Presentation:
- Evaluation.html
- Evaluation.qmd Fitted models are for FREE and for RENT now!!!!!! Just like Amazon has done for books and Apple has done for music. Just like Microsoft and Google have tried to do for software.
- Spotlight Software:
- Spotlight Software:
- Spotlight Software:
- Spotlight Software:
- Spotlight Software:
- Spotlight Blog:
- Spotlight Software:
- Spotlight Software:
- Spotlight Software:
- Spotlight Website:
- Spotlight Blog:
- An overview of gradient descent optimization algorithms List of the optimizers.
- Spotlight YouTube:
- keras:
Week 2:
- Handouts: Please open or print out the Handouts from Week 2 for this week.
- Spotlight Blog: Posit AI Blog
- Homework: Homework 2 has been posted.
- RStudio Keras Website: keras3
- Presentation:
- Presentation:
- Spotlight Paper: Leo Breiman Two Cultures
- Spotlight Wikipedia: Newton’s Method
- Spotlight Wikipedia: Newton’s Method for optimization
- Spotlight Wikipedia: Stochastic gradient descent
- Spotlight Blog: sentdex Difference between a Batch and an Epoch
- Code Example: We will look at the R code from from Chapter 4.
- Binary classification, movie_sentiment.qmd
- Multi-class classification, newswire_classification.qmd
- Regression, boston_housing.qmd
- Code Example: We will look at the R code from from Chapter 4. Colab Notebooks Being Reviewed keras3
- Binary classification, classifying-movie-reviews.R CoLab
- Multi-class classification, classifying-newswires.R CoLab
- Regression, predicting-house-prices.R CoLab
- NotebookLM: Stat654 Machine Learning Week01 notebook I uploaded the links to the presentations, R code, and the Zoom audio recordings to the notebook. I ask for an Audio Overview.
Week 1:
- Book: From the University Library > Databases A-Z > O > O’Reilly Then login using your university login. Find Deep Learning with R, Second Edition. Or the Python version - Book, audio book, or video book.
- Github: Download the code for the book from these links:
- Deep Learning with R Scripts and Python Notebooks with R code, Third Edition See code in the /code directory. The Second Edition code is is the /2e directory. We will be using the Third Edition code.
- Deep Learning with Python Notebooks, Third Edition
- Book: mdsr3e
- Book: r4ds2e
- Homework: Homework 1 has been posted.
- Presentation:
- Google Tensorflow Playground:
- playground
- deeperplayground
- playground.scienxlab
- TFPlaygroundPSA All of these datasets are simulated.
- Introductory Examples from my JSM 2018 Poster.
- R Project: Poster_vibe.zip
- cars Example
- concrete Example images.pdf
- Software Spotlight:
- Google Colab Instead of dealing with Java (on Windows you need the Windows Offline (64 bit) version) to run rWeka and h2o, you could use Google Colab. sign up for Pro as a Student FREE
- posit Cloud Instead of dealing with Java (on Windows you need the Windows Offline (64 bit) version) to run rWeka and h2o, you could use the posit Cloud.
- iris Example
- Poster:
- Spotlight Blog:
- NVIDIA CTG keynote: NVIDIA GTC 2026 Keynote updated
- NVIDIA CTG keynote: NVIDIA GTC 2025 Keynote
- Spotlight Book: Deep Learning and Scientific Computing with R torch
- Spotlight Book: Deep Learning
- Spotlight Book: Neural Networks and Deep Learning
- Spotlight Website: deeplearning.ai
- Spotlight Website: fast.ai
- Spotlight Website: Machine Learning Crash Course
- Google Learning:
- RStudio Keras Website: keras3
Week 0.5:
- YouTube:
- Run LLMs Locally:
- Links:
- Links: OpenAI
- Links: Microsoft
- bingChat aka CoPilot
- Microsoft CoPilot Designer
- Links: GitHub
- Links: Google
- Links:
- Links:
- Links:
Week 0:
Learning R:
Learning Python:
Learn SQL:
Excellent References:
Data Science:
- Socviz
- r4ds2e
- ModernDive
- Yarrr!
- R Data Science Essentials
- Python Data Science Essentials
- Deep Learning Made Easy with R
- Doing Data Science
- Data Science from Scratch
- What is Data Science? (fast easy read)
- Ethics and Data Science (fast easy read)
- Data Driven (fast easy read)
- R Markdown: The Definitive Guide
Reading related to the Digital Economy:
- The Second Machine Age: Work, Progress and Prosperity in a Time of Brilliant Technologies
- Race Against the Machine
- Wired For Innovation
- Strategies for e-business success
- Understanding the Digital Economy
More Big Picture:
- Fourth Paradigm of Science: Data-Intensive Scientific Discovery
- McKinsey Global Institute Big Data: The next frontier for innovation, competition, and productivity
Tech papers:
- Experimental evidence of massive-scale emotional contagion through social networks
- High-Resolution Image Synthesis with Latent Diffusion Models
- Attention Is All You Need
- A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivity
- Music can be reconstructed from human auditory cortex activity using nonlinear decoding models