vault backup: 2023-05-31 17:33:05

Affected files:
.obsidian/graph.json
.obsidian/workspace-mobile.json
.obsidian/workspace.json
Gaming/Steam controllers.md
History/Britain.md
STEM/AI/Neural Networks/CNN/CNN.md
STEM/AI/Neural Networks/CNN/FCN/FCN.md
STEM/AI/Neural Networks/CNN/FCN/ResNet.md
STEM/AI/Neural Networks/CV/Datasets.md
STEM/AI/Neural Networks/Properties+Capabilities.md
STEM/AI/Neural Networks/Transformers/Attention.md
STEM/AI/Properties.md
Tattoo/Engineering.md
Tattoo/Sources.md
Tattoo/img/snake-coil.png
Untitled.canvas

AI/Neural Networks/CNN/CNN.md

@@ -5,13 +5,13 @@
 - Niche
 	- No-one cared/knew about CNNs
 ## After
-- ImageNet
+- [[Datasets#ImageNet|ImageNet]]
 	- 16m images, 1000 classes
 - GPUs
 	- General processing GPUs
 	- CUDA
 - NIPS/ECCV 2012
-	- Double digit % gain on ImageNet accuracy
+	- Double digit % gain on [[Datasets#ImageNet|ImageNet]] accuracy
 
 # Full Connected
 [[MLP|Dense]]

AI/Neural Networks/CNN/FCN/FCN.md

@@ -5,7 +5,7 @@ Fully [[Convolution]]al Network
 
 Contractive → [[UpConv]]
 
-# Image Sementation
+# Image Segmentation
 - For visual output
 	- Previously image $\rightarrow$ vector
 - Additional layers to up-sample representation to an image

AI/Neural Networks/CNN/FCN/ResNet.md

@@ -23,6 +23,7 @@
 		- Except at end
 	- No dropout
 
+[[Datasets#ImageNet|ImageNet]] Error:
 ![[imagenet-error.png]]
 
 ![[resnet-arch.png]]

AI/Neural Networks/CV/Datasets.md

@@ -20,4 +20,9 @@
 	- Ship
 	- Truck
 - Achieved 90.7%
-	- Wan et al. 2013
+	- Wan et al. 2013
+
+# ImageNet
+- 14 million images
+- In at least one million of the images, bounding boxes are also provided
+- More than 20,000 categories

AI/Neural Networks/Properties+Capabilities.md

@@ -45,7 +45,7 @@
 	- Confidence value
 
 # Contextual Information
-- Knowledge represented by structure and activation weight
+- [[Neural Networks#Knowledge|Knowledge]] represented by structure and activation weight
 	- Any neuron can be affected by global activity
 - Contextual information handled naturally
 

AI/Neural Networks/Transformers/Attention.md

@@ -7,7 +7,7 @@
 	- Hyper-networks
 - Draw from relevant state at any preceding point along sequence
 	- Addresses [[RNN]]s vanishing gradient issues
-	- [[LSTM]] tends to poorly preserve far back knowledge
+	- [[LSTM]] tends to poorly preserve far back [[Neural Networks#Knowledge|knowledge]]
 - Attention layer access all previous states and weighs according to learned measure of relevance
 	- Allows referring arbitrarily far back to relevant tokens
 - Can be addd to [[RNN]]s

AI/Properties.md

@@ -1,7 +1,7 @@
 # Three Key Components
 
 1.  Representation
-	-   Declarative & Procedural knowledge
+	-   Declarative & Procedural [[Neural Networks#Knowledge|knowledge]]
 	-   Typically human-readable symbols
 2.  Reasoning
 	-   Ability to solve problems