This is an online glossary of terms for artificial intelligence, machine learning, computer vision, natural language processing, and statistics.
Useful Links
Terms
150 terms defined below, and 231 undefined terms.
- 1-bit Stochastic Gradient Descent (1-bit SGD)
- 1x1 Convolution
- Abstractive sentence summarization
- Activation function
- AdaGrad
- ADAM Optimizer
- Adaptive learning rate
- AlexNet
- Anchor box
- Autoencoder
- Bag-of-n-grams
- Bag-of-words
- Bagging
- Batch normalization
- Beam search
- Bias-variance tradeoff
- Bit transparency (audio)
- Boosting
- Bounding box
- Burstiness
- Capture-recapture model
- Catastrophic forgetting
- Chunking
- Co-adaptation
- Codebook
- Codebook collapse
- Computer vision
- Connectionist Temporal Classification (CTC)
- Convex combination
- Convex hull
- Convolution
- Coreference resolution
- Cross validation
- Data augmentation
- Data parallelism
- Deep Convolutional Generative Adversarial Network (DCGAN)
- Dimensionality reduction
- Distance metric
- Distance metric learning
- Distributed representation
- Distributional similarity
- doc2vec
- Dropout
- Dying ReLU
- Embed, Encode, Attend, Predict
- Entailment
- Extractive sentence summarization
- Face detection
- Face recognition
- Face verification
- Facet
- Facet (disambiguation)
- Facet (visualization tool)
- fastText
- Filter (convolution)
- First-order information
- Generative Adversarial Network (GAN)
- Gradient
- Gradient descent
- Gram matrix
- Hadamard product
- Harmonic Precision-Recall Mean (F1 Score)
- He initialization
- Hinge loss
- Hopfield Network
- Imputation
- Inceptionism
- Independent and Identically Distributed (i.i.d)
- Internal covariate shift
- Inverted dropout
- Jaccard index (intersection over union)
- Language segmentation
- Latent Dirichlet Allocation Differential Evolution (LDADE)
- Leaky ReLU
- LeNet
- Long Short-Term Memory (LSTM)
- Magnet loss
- Margin
- Mean Reciprocal Rank (MRR)
- METEOR
- Minimax
- Mode collapse
- Model parallelism
- Momentum
- Multinomial mixture model
- Multiple crops at test time
- Named Entity Recognition in Query (NERQ)
- NCHW
- Neural checklist model
- Neural style transfer
- Neural Turing Machine (NTM)
- NHWC
- Non-max suppression
- Object localization
- One-hot encoding
- One-shot learning
- Out-of-core
- Overfitting
- Padding (convolution)
- Paragraph vector
- Parameter budget
- Part-of-Speech (POS) Tagging
- PCA Color Augmentation
- Peephole connection
- Perplexity
- Pooling layer
- Prompt engineering
- Provenance tracking
- Pseudo-labeling
- Random initialization
- Rectified Linear Unit (ReLU)
- Recurrent Neural Network
- Reinforcement Learning (RL)
- ResNet
- Same convolution
- Search-based software engineering (SBSE)
- Self-supervised learning
- Semantic hashing
- Semi-supervised learning
- sense2vec
- Sequence to Sequence Learning (seq2seq)
- Siamese neural network
- Similarity learning
- Sobel filter (convolution)
- Softmax
- Stationary environment
- Supervised learning
- Support Vector Machine (SVM)
- Symmetry breaking
- Synset
- SyntaxNet
- Termite
- Top-1 error rate
- Top-5 error rate
- Training, validation, and test datasets
- Tree-LSTM
- Triplet loss
- UNK
- Valid convolution
- Variance
- Vector-Quantized Variational Autoencoders (VQ-VAE)
- VGGNet
- Weak supervision
- Weight sharing
- Word embedding
- word2phrase
- word2vec
- YOLO (object detection algorithm)
- Zero padding
- Zero-shot learning