vault backup: 2023-06-02 17:17:29

Affected files:
.obsidian/workspace-mobile.json
.obsidian/workspace.json
Lab/Scratch Domain.md
Money/Econ.md
STEM/AI/Classification/Classification.md
STEM/AI/Classification/README.md
STEM/AI/Classification/Supervised.md
STEM/AI/Neural Networks/CNN/Examples.md
STEM/AI/Neural Networks/CNN/FCN/FCN.md
STEM/AI/Neural Networks/CNN/FCN/FlowNet.md
STEM/AI/Neural Networks/CV/Filters.md
STEM/img/coordinate-change.png
STEM/img/gaussian-class.png
Tattoo/Engineering.md
Want.md

AI/Classification/Classification.md

@@ -0,0 +1,34 @@
+*Given an observation, determine one class from a set of classes that best explains the observation*
+
+***Features are discrete or continuous***
+
+- 2 category classifier
+	- Dichotomiser
+
+# Argmax
+Argument that gives the maximum value from a target function
+
+# Gaussian Classifier
+[Training](Supervised.md)
+- Each class $i$ has it's own Gaussian $N_i=N(m_i,v_i)$
+
+$$\hat i=\text{argmax}_i\left(p(o_t|N_i)\cdot P(N_i)\right)$$
+$$\hat i=\text{argmax}_i\left(p(o_t|N_i)\right)$$
+- With equal priors
+
+![](../../img/gaussian-class.png)
+
+# Discrete Classifier
+- Each class $i$ has it's own histogram $H_i$
+	- Describes the probability of each observation type $k$
+	- $P(o_t=k|H_i)$, based on class-specific type counts
+
+$$\hat i=\text{argmax}_i\left(P(o_t=k|H_i)\right)$$
+- Nothing else known about classes
+
+$$\hat i=\text{argmax}_i\left(P(o_t=k|H_i)\cdot P(H_i)\right)$$
+- Given class priors $P(H_i)$
+- Maximum posterior probability
+	- Bayes
+
+![](../../img/coordinate-change.png)

AI/Classification/README.md

@@ -0,0 +1 @@
+Classification.md

AI/Classification/Supervised.md

@@ -0,0 +1,5 @@
+
+# Gaussian Classifier
+- With $T$ labelled data
+
+$$q_t(i)=$$

AI/Neural Networks/CNN/Examples.md

@@ -33,7 +33,7 @@
 
 ![googlenet](../../../img/googlenet.png)
 
-## [[Inception Layer]]
+## [Inception Layer](Inception%20Layer.md)
 ![googlenet-inception](../../../img/googlenet-inception.png)
 ## Auxiliary [[Deep Learning#Loss Function|Loss]] Functions
 - Two other SoftMax blocks

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

@@ -3,7 +3,7 @@ Fully [[Convolution]]al Network
 [[Convolutional Layer|Convolutional]] and [[UpConv|up-convolutional layers]] with [[Activation Functions#ReLu|ReLu]] but no others (pooling)
 - All some sort of Encoder-Decoder
 
-Contractive → [[UpConv]]
+Contractive → [UpConv](../UpConv.md)
 
 # Image Segmentation
 - For visual output
@@ -29,7 +29,7 @@ Contractive → [[UpConv]]
 - Can use summed per-pixel log [[Deep Learning#Loss Function|loss]]
 
 # Evaluation
-![[fcn-eval.png]]
+![fcn-eval](../../../../img/fcn-eval.png)
 - SDS
 	- Classical method
 	- 52% mAP

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

@@ -3,9 +3,9 @@ Optical Flow
 - 2-Channel optical flow
 	- $dx,dy$
 - Two consecutive frames
-	- 6-channel [[tensor]]
+	- 6-channel [tensor](../../../../Maths/Tensor.md)
 
-![[flownet.png]]
+![flownet](../../../../img/flownet.png)
 
 # [[Skip Connections]]
 - Further through the network information is condensed
@@ -14,13 +14,13 @@ Optical Flow
 	- Append activation maps from encoder to decoder
 
 # Encode
-![[flownet-encode.png]]
+![flownet-encode](../../../../img/flownet-encode.png)
 
 # [[Upconv]]
-![[flownet-upconv.png]]
+![flownet-upconv](../../../../img/flownet-upconv.png)
 
 # Training
 - Synthetic rendered objects
 - Real background images
 
-![[flownet-training.png]]
+![flownet-training](../../../../img/flownet-training.png)

AI/Neural Networks/CV/Filters.md

@@ -1,2 +1,2 @@
 # Gabor
-![[gabor.png]]
+![gabor](../../../img/gabor.png)