---
title: "Biostat M280 Homework 4"
subtitle: Due Mar 22 @ 11:59PM
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

For this homework, you can work as a team of size $\le 5$. You can create a new private GitHub repository for collaboration (need to add @Hua-Zhou and @juhkim111 as collaborators) or re-use the current repository of a team representative. For each question, your report should have a clear description of role of each team member, and Git log should reflect individual contribution to the project.

## Q1 Learn by doing

I found the [TensorFlow for R Blog](https://blogs.rstudio.com/tensorflow/) series at RStudio quite illuminating. Choose one blog that interests you and do following.  

1. Reproduce the results in the blog.  

2. Make your own tweaks. For example, try different tuning parameter values and report what you found, or try a new data set, or apply the method to a new application.

## Q2 Deep learning on smart phone

Professor May Wang in Department of Community Health Sciences (CHS) studies obesity in children and intervention strategies to prevent obesity. She asked me whether it is possible to develop an app such that a user takes a photo of a meal and the app will recognize and record the type of food (pizza, mac and cheese, burger, ...). 

Your job: produce a prototype app for iPhone or Android smart phone. 

Resources:  
1. There are plenty of tutorials and YouTube clips on making apps for iPhone or Android.  
2. Google's [Cloud Vision API](https://cloud.google.com/vision/) may supply an easy cloud solution.  
3. [TensorFlow Lite](https://www.tensorflow.org/lite) may provide an easy mobile solution.