Weekly Update for August 6

class: center, middle, inverse, title-slide

# Weekly Update for August 6
### 2018/08/06

---

class: inverse
# About Slides

---
class: primary 
# New CSAFE slide template

We're now using [`xaringan`](https://github.com/yihui/xaringan)

What's changed:

- New person slide:

````
---
class: inverse
# Your Name
````

---
class: secondary

- New content slide with title:

````
---
class: primary
# Title of slide

Slide content
````

- New content slide without title:

````
---
class: secondary

Slide content with no title on slide
````

---
class: inverse
# Sample User

---
class: primary
# Sample Slide

- Sample Table (top press freedom index):

| Country      | 2018 Press Freedom Index |
| :----------- | :----------------------- |
| Norway       | 7.63   |
| Sweden       | 8.31   |
| Netherlands  | 10.01  |
| Finland      | 10.26  |
| Switzerland  | 11.27  |
| Jamaica      | 11.33  |

---
class: secondary

- If you are going to add an image, **create a directory** with your name within "images" folder. For example, "images/**guillermo**/sample_image.JPG"

<img src = "images/guillermo/sample_image.JPG" width="35%"/>
 
---
class: inverse 
# Nate
---
class: primary
# Groove ID

- Checked the results of the changepoint algorithm on about 40 lands 
- Takeaways so far: 
    - seems to work well when there are two shoulders in an image
    - multimodal posterior is still an issue
    - it can be easy to fool the method to think there are shoulders when none exist
- Would like to try adding a linear function of the location to the mean of the GPs assigned to the shoulders

---
class: primary
# SLR

- Working on simulations to verify the legitimacy of some simple derivations
- May be some justifcation based on CLT for assuming that the difference between log(LR) and log(SLR) is approximately normal if:
    - original data is truly high dimensional
    - the score function is a kind of distance
- If original data is Gaussian, the bias of the log(SLR) is a function of the ratio of the variances of source and trace data

---
class: inverse
# Amy

---
class: primary
# Updates from MN

Hi all! I'm working from MN on the following:
1. Handling 'sticky sampling' issues with latent variables in an extension of the handwriting model
2. Formal model selection methods for the number of graphemes to include in modeling (DIC is giving trouble)
3. Incorporating a parameter that captures document to document variability within writer
4. Data collection: IRB submission by August 21

---
class: inverse
# Jimmy

---
class: primary
# Last Few Weeks

* 2 procedure types created (YES students)

* IAI
  + Great response about CSAFE
  + Interest in the Webinars 
  + FBI is interested in creating a shoe data base
  + Talks on 3D Scanning, latent print analysis, and a case study

<img src = "images/Jim/star.png" width="50%"/>
---
class: primary
# Surveys and This Week

* Roughly 480 surveys

* Re-entering and reviewing (Missing Surveys???)

* Cleaning and attempting to make the data uniform.

* Meeting with Eric tomorrow

---
  class:inverse
#Miranda

---
  class:primary
# Model Predictions

* With new GPU server, training goes from 28 hours to 1.5
* Trained 3-class model with single label
* Pictured: images of "triangles", but model detects other shapes in images

---
class: inverse
# Susan

---
class: primary
# Currently working on...

- bulletxtrctr tests
  - style checks
  - adding s3 classes?
- Checking shoe annotations for accuracy/completeness
- Working through zotero setup - CSAFE citation/pdf library

---
class: inverse
# Earo

---
class: primary
# $whoami

* third-year Ph.D. @ Monash Uni

* R developer, authoring `tsibble`, `sugrrants`, `hts`, and etc.

* more? <https://earo.me>

---
class: secondary