2021-03-30-Tue-Study
Mar 30, 2021
»
study
RL Class Review
Supervised Learning:
Linear Regression
A linear model that establishes the relationship btw a dependent variable y(target)
and one or more independent variables denoted x (inputs)
*Goal is to find a straight line that is best fit
Could be determined by statistics(LSM) and Gradient Descent(optimization theory)
Gradient Descent equation
y = nx + b
b = weight
error = (predicted y - actual y)+
Cost Function
Gradient Descent Procedure
1. draws the line (tan) from the point
2. finds the slope of that line
3. identifies how much change is require by
taking parital derivative of the function with the respective to θ
*partial derivative = (d/dθ)
4. change value will be multiplied with learning rate called α
5. To get new θ = θ - change value*α
Unsupervised learning
System has to understand itself from the data given
ex) Clustering
Task of dividing the population or data points into a number of groups
such that data points in the same groups are more similar to other data points
in the same group and dissimilar to the data points in other groups
Deep RL
Standard reinforcement learning where a deep neural network is used
to approximate either a policy or a value function
Deep Neural networks require lots of real interaction with the environment
to learn can easily parallelize the interaction
Python
mutable = Value Changes
Number
String
Dictionary = {:}
immutable = Value unchanges
List = []
Tuple = ()
Linux
Hard Link
Mirror copy of the original file
Even if you delete the original file, the hard link still
points to same inode. So it works
ln [original file name][hard link name]
Soft Link
Actual link to the original file
If you delete the original file, the soft link has no value
becauseit points to a non-existent file
ln -s [original file name][soft link name]