AI & ML Glossary

Terms from the courses, in one place. Search for what you need, and tap + Flashcard on anything you want to remember. It'll come back in your daily review until it sticks.

A

a
A neuron's activation: the output after applying a non-linear activation function to z. Deep Learning · Neurons & Activations · view in course
Adjusted R²
R² penalized for the number of features used, so it doesn't automatically rise when you add useless features. Machine Learning Fundamentals · Evaluation Metrics · view in course
α (alpha)
The regularization strength in ridge, lasso, and Elastic Net. Larger α means a stronger penalty on coefficient size. Machine Learning Fundamentals · Linear Models · view in course
Attention weights
A softmax distribution over input positions, derived from query/key similarity, used to weight a combination of values. Deep Learning · Attention & Transformers · view in course

B

b0, b1, ..., bn (or β0, β1, ..., βn)
Intercept and per-feature coefficients in multiple linear regression: y = b0 + b1·x1 + ... + bn·xn. Machine Learning Fundamentals · Linear Models · view in course
Backpropagation through time (BPTT)
Backpropagation applied to an unrolled recurrent network, propagating gradients backward across time steps as well as layers. Deep Learning · Forward & Backward Propagation · view in course
Backpropagation
Computing the gradient of the loss with respect to every weight by applying the chain rule backward from the output layer to the input layer. Deep Learning · Forward & Backward Propagation · view in course
Bagging
Bootstrap aggregating: training many models on random subsamples of the data (drawn with replacement) and averaging their predictions. The basis of random forests. Machine Learning Fundamentals · Trees & Ensembles · view in course
Batch normalization
Normalizing activations within each mini-batch to stable mean and variance, reducing internal shift between layers and stabilizing training. Deep Learning · Regularization · view in course
Batch size
The number of training examples used to compute one gradient update. Deep Learning · Practical Training · view in course
Bayes' theorem
P(class | features) = P(features | class) · P(class) / P(features). Lets you flip a conditional probability around. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course
Bias
Error from a model being too simple to capture the true pattern. Systematic and repeatable, regardless of which training sample it saw. Machine Learning Fundamentals · Core Concepts · view in course
Boosting
Training models sequentially, where each new model corrects the residual errors of the ensemble built so far. The basis of gradient boosting and XGBoost. Machine Learning Fundamentals · Trees & Ensembles · view in course

C

C (SVM)
Controls how heavily margin violations are penalized. Large C means a narrower, stricter margin; small C means a wider, more tolerant one. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course
Cell state (LSTM)
A separate memory channel in an LSTM, regulated by gates, that lets information persist over much longer sequences than a vanilla RNN's hidden state alone. Deep Learning · Sequence Models · view in course
Centroid
The mean position of all points currently assigned to a cluster in K-means. Machine Learning Fundamentals · Unsupervised Learning · view in course
Conditional independence (Naive Bayes)
The assumption that every feature is independent of every other feature, given the class. Rarely exactly true, but often harmless for ranking classes. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course
Cross-entropy loss
The standard classification loss, paired with a softmax output. Its gradient stays informative even for confidently wrong predictions. Deep Learning · Forward & Backward Propagation · view in course
Curse of dimensionality
As feature count grows, distances between points become less meaningful, weakening any purely distance-based method. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course

D

Data augmentation
Applying realistic transformations (crops, flips, rotations, etc.) to training examples to expose the model to more variation without collecting new data. Deep Learning · Regularization · view in course
Dead neuron
A ReLU neuron stuck outputting zero for every input, because its weights drove it permanently into the negative, zero-gradient region. Deep Learning · Neurons & Activations · view in course
Dropout
Randomly zeroing a fraction of activations on each forward pass during training, so the network can't over-rely on any single neuron. Deep Learning · Regularization · view in course

E

Early stopping
Halting training once validation performance stops improving, keeping the weights from before the model began overfitting. Deep Learning · Regularization · view in course
Encoder-decoder
An architecture where an encoder compresses an input sequence into a representation that a decoder uses to generate an output sequence, possibly of a different length. Deep Learning · Sequence Models · view in course
Entropy
−Σ pᵢ log₂(pᵢ). An information-theoretic alternative to Gini impurity, used to compute information gain. Machine Learning Fundamentals · Trees & Ensembles · view in course
Epoch
One full pass through the entire training dataset. Deep Learning · Practical Training · view in course
Explained variance ratio
The fraction of a dataset's total variance a single principal component captures on its own. Machine Learning Fundamentals · Unsupervised Learning · view in course
Exploding gradient
Gradients growing without bound as they're propagated backward, causing unstable, divergent updates. Deep Learning · Forward & Backward Propagation · view in course

F

f / ŷ ("y-hat")
The model's prediction. Machine Learning Fundamentals · Core Concepts · view in course
F1 score
The harmonic mean of precision and recall: 2 · (Precision · Recall) / (Precision + Recall). High only when both are reasonably high. Machine Learning Fundamentals · Evaluation Metrics · view in course
Feature map
The grid of outputs produced by sliding one filter across an input; high values mark where that filter's pattern was detected. Deep Learning · Convolutional Networks · view in course
Fine-tuning
Continuing to train a pretrained model's weights (often with a small learning rate, sometimes with early layers frozen) on a new, usually smaller, task-specific dataset. Deep Learning · Attention & Transformers · view in course
Forget / input / output gate
The three learned gates in an LSTM that control what is discarded from, added to, and read out of the cell state at each time step. Deep Learning · Sequence Models · view in course
Forward propagation
Computing a network's output by passing an input through each layer's weights, bias, and activation in sequence. Deep Learning · Forward & Backward Propagation · view in course

G

Gini impurity
1 − Σ pᵢ². Roughly the probability that two randomly chosen points from a group have different labels. 0 means perfectly pure. Machine Learning Fundamentals · Trees & Ensembles · view in course
GRU
Gated Recurrent Unit: merges the cell and hidden state into one and uses two gates instead of three, a simpler alternative to the LSTM. Deep Learning · Sequence Models · view in course

H

H
The hypothesis space: the family of functions an algorithm is allowed to search through. Training means finding the f ∈ H that makes f(x) as close to y as possible. Machine Learning Fundamentals · Core Concepts · view in course
He initialization
Like Xavier, but scaled up to compensate for ReLU zeroing out roughly half its inputs. Deep Learning · Optimization · view in course
Hidden state
The vector an RNN carries forward between time steps, summarizing everything the network has processed in the sequence so far. Deep Learning · Sequence Models · view in course

I

Inertia
The total squared distance from every point to its assigned centroid. Always decreases as K increases; the elbow method looks for where it stops decreasing fast. Machine Learning Fundamentals · Unsupervised Learning · view in course

K

K (K-means)
The number of clusters K-means is told to find. Chosen up front, often via the elbow method. Machine Learning Fundamentals · Unsupervised Learning · view in course
K (KNN)
The number of nearest neighbors consulted when classifying a new point. Small K means high variance; large K means high bias. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course
K-fold cross-validation
Rotating which portion of the data is used for validation across K rounds, so every sample contributes to both training and validation without ever doing both at once. Machine Learning Fundamentals · Model Quality · view in course
Kernel / filter
A small grid of learned weights that slides across an input, computing a weighted sum of each local patch it covers. Deep Learning · Convolutional Networks · view in course
Kernel trick
Computing what a dot product would be in a higher-dimensional space, without explicitly building that space, letting SVM handle non-linear boundaries. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course

L

L1 penalty (lasso)
Penalizes the sum of absolute coefficient values. Can shrink coefficients to exactly zero, which performs feature selection. Machine Learning Fundamentals · Linear Models · view in course
l1_ratio
In Elastic Net, the mix between L1 and L2 penalty, from 0 (pure ridge) to 1 (pure lasso). Machine Learning Fundamentals · Linear Models · view in course
L2 penalty (ridge)
Penalizes the sum of squared coefficients. Shrinks every coefficient toward zero smoothly, but rarely to exactly zero. Machine Learning Fundamentals · Linear Models · view in course
Learning rate (η)
The step size used in a gradient descent update: w ← w − η·∇w. Deep Learning · Optimization · view in course
Learning rate decay
Shrinking the learning rate over the course of training, taking large early steps and smaller, more precise steps later. Deep Learning · Optimization · view in course
Loss function
A number measuring how wrong a model's predictions are. Training minimizes this number. Machine Learning Fundamentals · Core Concepts · view in course

M

m, b
Slope and intercept of a simple linear regression line: y = mx + b. Machine Learning Fundamentals · Linear Models · view in course
MAE
Mean Absolute Error: the average of |actual − predicted|. Weighs every error linearly. Machine Learning Fundamentals · Evaluation Metrics · view in course
Margin (SVM)
The distance between a decision boundary and the closest training points on either side. SVM maximizes this. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course
Mixed precision
Using lower-precision number formats (e.g. float16) for most computation, combined with float32 for numerically sensitive steps, to speed up training. Deep Learning · Practical Training · view in course
Momentum
A running average of past gradients added to the current update, smoothing noisy steps and accelerating progress in consistent directions. Deep Learning · Optimization · view in course
MSE / RMSE
Mean Squared Error and its square root. Squares errors before averaging, so large errors count disproportionately more. Machine Learning Fundamentals · Evaluation Metrics · view in course
Multi-head attention
Running several attention computations in parallel, each with its own learned projections, so different heads can specialize in different relationships. Deep Learning · Attention & Transformers · view in course
Multicollinearity
When two or more features are highly correlated, making individual coefficients unstable to interpret even though overall predictions stay accurate. Machine Learning Fundamentals · Linear Models · view in course

N

n_estimators, learning_rate, max_depth
The three core gradient-boosting hyperparameters: number of trees, how much each tree's correction counts, and how deep each tree is allowed to grow. Machine Learning Fundamentals · Trees & Ensembles · view in course

O

Overfit-a-tiny-batch test
A debugging sanity check: a correctly wired model should be able to drive its loss near zero on a handful of examples it should easily memorize. Deep Learning · Practical Training · view in course
Overfitting
Training error keeps falling while validation error rises. The model has started fitting noise instead of signal. Machine Learning Fundamentals · Model Quality · view in course

P

Padding
Adding a border (usually zeros) around an input before convolving, used to control the output's spatial size. Deep Learning · Convolutional Networks · view in course
Pooling (max / average)
Downsampling a feature map by keeping only the max (or average) value within each small local region. Deep Learning · Convolutional Networks · view in course
Positional encoding
A signal added to input embeddings that injects information about token order, since self-attention has no inherent sense of position. Deep Learning · Attention & Transformers · view in course
Precision
TP / (TP + FP). Of everything flagged positive, how much actually was. Machine Learning Fundamentals · Evaluation Metrics · view in course
Principal component
A new axis found by PCA, aligned with a direction of variance in the data. Ordered from most to least variance captured. Machine Learning Fundamentals · Unsupervised Learning · view in course

Q

Query, key, value (Q, K, V)
The three projections of an input used in attention: a query is compared against keys to compute weights, which combine the values. Deep Learning · Attention & Transformers · view in course

R

Coefficient of determination: 1 − (Sum of Squared Residuals / Total Sum of Squares). The fraction of variance in the target explained by the model, relative to always guessing the mean. Machine Learning Fundamentals · Evaluation Metrics · view in course
Recall
TP / (TP + FN). Of everything actually positive, how much was caught. Machine Learning Fundamentals · Evaluation Metrics · view in course
ReLU
max(0, z). The default hidden-layer activation in most modern networks; cheap to compute and avoids vanishing gradients for positive inputs. Deep Learning · Neurons & Activations · view in course
Residual / skip connection
A shortcut that adds a layer's input directly to its output, letting gradients flow around the layer and making very deep networks easier to train. Deep Learning · Convolutional Networks · view in course
Residual
The gap between an actual value and a model's predicted value: actual − predicted. Machine Learning Fundamentals · Linear Models · view in course
ρ (rho)
The correlation between trees in an ensemble. The variance of an average of n trees is ρσ² + (1−ρ)σ²/n, so ρσ² is a variance floor more trees alone can't beat. Machine Learning Fundamentals · Trees & Ensembles · view in course

S

Self-attention
Attention where queries, keys, and values all come from the same sequence, letting every position relate to every other position. Deep Learning · Attention & Transformers · view in course
SGD
Stochastic gradient descent: updating weights using the gradient from one example or a small mini-batch, rather than the full dataset. Deep Learning · Optimization · view in course
Sigmoid / tanh
Smooth, saturating activations bounded in (0, 1) or (−1, 1). Prone to vanishing gradients in deep networks since their derivative shrinks toward zero at the extremes. Deep Learning · Neurons & Activations · view in course
Softmax
Converts a vector of raw scores (logits) into a probability distribution: exponentiate each value, then normalize so they sum to 1. Deep Learning · Forward & Backward Propagation · view in course
Stride
The step size a filter moves between positions. Larger strides produce smaller output feature maps. Deep Learning · Convolutional Networks · view in course
Support vectors
The training points closest to the decision boundary. They're the only points that determine where the boundary sits. Machine Learning Fundamentals · Distance-Based & Margin-Based Methods · view in course

T

TP, FP, TN, FN
True/false positive, true/false negative: the four cells of a confusion matrix. Machine Learning Fundamentals · Evaluation Metrics · view in course
Train / validation / test split
Three distinct data subsets: one to fit parameters, one to make decisions about the model, and one held back untouched for a single final honest check. Machine Learning Fundamentals · Model Quality · view in course

U

Underfitting
Both training and validation error are high and close together. The model is too constrained to capture the true pattern. Machine Learning Fundamentals · Model Quality · view in course
Unrolling
Representing an RNN's repeated application across time steps as one long computational graph with shared weights, so backpropagation can be applied across the whole sequence. Deep Learning · Sequence Models · view in course

V

Vanishing gradient
Gradients shrinking toward zero as they're propagated backward across many layers or time steps, stalling learning in early layers. Deep Learning · Forward & Backward Propagation · view in course
Variance
Error from a model being too sensitive to the specific training sample it happened to see. Retrain on different data, get a noticeably different model. Machine Learning Fundamentals · Core Concepts · view in course

W

w, b
A neuron's weights and bias. Its pre-activation output is z = w·x + b. Deep Learning · Neurons & Activations · view in course
Warmup
Starting training with a small learning rate and gradually increasing it, to avoid destabilizing updates while weights are still freshly initialized. Deep Learning · Optimization · view in course

X

x
An input: a single example's features. Machine Learning Fundamentals · Core Concepts · view in course
Xavier / Glorot initialization
Scales initial random weights based on layer size to keep activation variance roughly stable across layers, suited to symmetric activations like tanh. Deep Learning · Optimization · view in course

Y

y
The true label or target value for an example. Machine Learning Fundamentals · Core Concepts · view in course

Z

z
A neuron's pre-activation value, before an activation function is applied. Deep Learning · Neurons & Activations · view in course