commit ce90e5ac9a6e8ec49569e31a83d4650491c164e7 parent e7816522f701354b60125d78469103c62086b736 Author: Andrew <andrewlaack1@gmail.com> Date: Thu, 14 Nov 2024 22:00:12 -0600 Took some other notes Diffstat:
| M | MachineLearning.md | | | 26 | +++++++++++++++++++++++--- |
| M | ReinforcementLearning.md | | | 3 | +++ |
| M | Weight.md | | | 4 | +++- |
3 files changed, 29 insertions(+), 4 deletions(-)
diff --git a/MachineLearning.md b/MachineLearning.md @@ -32,14 +32,34 @@ h(x) = this is the function with an input of x this should be about the correct Deep Learning With Python (Francois Chollet): -Ch 1: +Ch 1 (What is DL): * [RepresentationLearning](RepresentationLearning.md) * [LossFunction](LossFunction.md) * [UtilityFunction](UtilityFunction.md) -Ch 2: - +Ch 2 (Maths behind DL): + +* Representation +* DataDistillation +* Softmax +* [Optimizer](Optimizer.md) +* Tensor - Dimension = Axis +* TensorSlicing - Select specific element +* BatchAxis - Batch Dimension +* Rank - axis count of tensor +* Scalar - OD tensor +* Overfitting +* Broadcast - match lower dim tensor with higher generally for element wise comparison +* [Transpose](Transpose.md) +* AffineTransformation - Linear transformation + Tranaslation (note that affines composed is still simply an affine thus we need activation functions) +* HypothesisSpace +* GeometricTransformation +* Manifold +* Kernel (multiply part of weights) +* [Bias](Bias.md) +* [Weight](Weight.md) +* Surface ISL Python: diff --git a/ReinforcementLearning.md b/ReinforcementLearning.md @@ -46,3 +46,6 @@ L4 * [MonteCarloLearning](MonteCarloLearning.md) * [IncrementalMean](IncrementalMean.md) * [TemporalDifferenceLearning](TemporalDifferenceLearning.md) + +L5 + diff --git a/Weight.md b/Weight.md @@ -5,4 +5,6 @@ ML D6 ## Notes -**Definition:** Weights in ANNs are numerical values that represent the strength of connections between neurons (perceptrons). +**Definition:** Weights in ANNs are numerical values that represent the strength of connections between neurons and biases. + +The connection strengths are called kernels and the sum of these + biases is the total number of trainable parameters (weights).