https://www.calledtocommunion.com/2013/11/lawrence-feingold-the-motives-of-credibility-for-faith/

- A Markov process has the Markov Property
- Action Policy is a function that returns an action given an state
- An AI Agent is an entity that can observe and act autonomously
- Assistolia Fetal
- Avoid python loops when Pytorch functions are available
- Bellman Equation express a value function
- Building a Linear Regression Model
- Classification vs Regression models
- Convolutional Neural Networks
- Data sets divide input data
- Deep Learning Jargon
- Deep Q Network (DQN) uses deep learning to do Q-learning
- Discount rates reflect timing and risk
- Discounted Cash Flows is a way of valuing an asset
- DQN training doesn't rely on expected outputs
- Epoch is a training cycle in model training
- Epsilon-greedy dictates the action of the agent during training
- Fine-tuning
- Full text-search
- Gell-Mann Amnesia effect
- Good code has good comments
- Hash Collision
- Hashcash is a proof of work algorithm
- How can machines learn?
- How to sync Obsidian iOS with Google Drive
- Information Hiding
- Lectio Divina
- Loss functions
- Markov chain is a markov process
- Markov Decision Process (MDP) is influenced by decisions and stochasticity
- Markovian text generation
- Matrix multiplication
- Moravec's Paradox
- Motives of Credibility
- Multi-writes may cause a system to transition to an invalid state
- Mutations are not random
- O uso de cloreto de potássio para eutanásia em animais é inaceitável no Brasil.
- One dollar today is more valuable than one dollar tomorrow
- Operations with tensors of different ranks use broadcasting
- Overfitting is a failure of model training
- Pre-trained model
- Pytorch tensor is a fast array for GPUs
- Q-function is a value function approximation
- Q-learning is the process of finding the best possible Q-function
- Quality metrics for estimators models
- Regression Analysis
- Reinforcement Learning
- ReLUs are used to introduce non-linearity in neural networks
- Replay Buffer register what is happening
- Residuals is the difference between the predicted value and actual value
- Sessions vs JWTs
- Sigmoids are used to ensure the value is between 0 and 1
- State Machines
- State Transition Matrix maps states transitions according to actions.
- Stochasticity
- The head is the top of the neural network
- Train-Test Split
- Transfer learning
- Weights and biases are the paremeters of a neural network
- Why the earlier layers don't require many changes in fine-tuning

https://www.calledtocommunion.com/2013/11/lawrence-feingold-the-motives-of-credibility-for-faith/

- No backlinks found