From ebb247a5b908687deab5dcdb0767890befc3f5cc Mon Sep 17 00:00:00 2001
From: Jet Hughes <jethughes0@gmail.com>
Date: Mon, 10 Oct 2022 10:12:45 +1300
Subject: [PATCH] vault backup: 2022-10-10 10:12:45

---
 content/notes/18-ML-in-IA-2.md | 44 +++++++++++++++++++++++++++++++++-
 content/notes/19-deepfakes.md  | 18 ++++++++++++++
 content/notes/comp-210.md      |  1 +
 3 files changed, 62 insertions(+), 1 deletion(-)
 create mode 100644 content/notes/19-deepfakes.md

diff --git a/content/notes/18-ML-in-IA-2.md b/content/notes/18-ML-in-IA-2.md
index 00709d23e..148f3cbc5 100644
--- a/content/notes/18-ML-in-IA-2.md
+++ b/content/notes/18-ML-in-IA-2.md
@@ -4,6 +4,9 @@ aliases:
 tags: 
 - lecture
 - comp210
+sr-due: 2022-10-13
+sr-interval: 3
+sr-ease: 250
 ---
 
 # nefarious uses of ml
@@ -71,9 +74,48 @@ how
 
 F5 steganographic algorithm
 - developed to fool analysis of dct distributions
-- 
+- seeded with key to create pseudorandom sequence for embedding
+- can preserve statistical properties of DCT distributions
 
+can use ML to find hidden images
+- then hackers will try to fool this
+- some will always get through
 
 # bigger issues
+- deepfakes to to shape political views of the day
+- pixel replacement with segmentation and inpainting 
+	- ![examples](https://i.imgur.com/zGOtqZa.png)
+
+## is ML good or bad
+- being used everywhere
+
+- should we care
+
+- data and modelling cannot always be 100% perfect
+	- e.g., killer drones
+
+- privacy concerns
+- linked data
+- pipelins - information seepage
+
+nx integrated data infrastructure
+
+ethics
+- what considerations need to be made
+- ML being used to automate decision making
+- ML sentencing of criminals
+
+theft
+- theft of data
+- data is more valuable
+- transfer learning
+
+- reverse engineering a ML model
+![ml extraction attack](https://i.imgur.com/jiinX6m.png)
+
 
 # where to from here
+- good and bad are human constructs
+- how will laws work
+- can we use ML to make laws
+- Do we need to stop it?
\ No newline at end of file
diff --git a/content/notes/19-deepfakes.md b/content/notes/19-deepfakes.md
new file mode 100644
index 000000000..5a95ef5a6
--- /dev/null
+++ b/content/notes/19-deepfakes.md
@@ -0,0 +1,18 @@
+---
+title: "19-deepfakes"
+aliases: 
+tags: 
+- comp210
+- lecture
+---
+
+# what, origins, how
+- issue with trust politics etc
+
+- 
+
+# related concepts
+
+# how
+
+# concerns and solutions
diff --git a/content/notes/comp-210.md b/content/notes/comp-210.md
index 6776b3e3e..e5745dc26 100644
--- a/content/notes/comp-210.md
+++ b/content/notes/comp-210.md
@@ -38,4 +38,5 @@ No final exam
 - [15-policies-standards-practices](notes/15-policies-standards-practices.md)
 - [16-cloud-security](notes/16-cloud-security.md)
 - [17-ML-in-IA-1](notes/17-ML-in-IA-1.md)
+- [18-ML-in-IA-2](notes/18-ML-in-IA-2.md)
 - 
\ No newline at end of file