Primers • AI

Overview

• Here’s a hand-picked selection of articles on AI fundamentals/concepts that cover the entire process of building neural nets to training them to evaluating results.

Design/Architecture

• ML Algorithms Comparative Analysis
• DL Architectures Comparative Analysis
• Generative Adversarial Networks (GANs)
• Diffusion Models
• Graph Neural Networks
• Attention
• Separable Convolutions
• Inductive Bias
• Multiclass vs. Multilabel Classification

Training

• Xavier Initialization
• Padding
• Regularization
• Gradient Descent and Backprop
• Activation Functions
• Loss Functions
• Fine-tuning Models
• Splitting Datasets
• Batchnorm
• Dropout
• Double Descent
• Fine-Tuning and Evaluating BERT
• Debugging Deep Learning Projects
• Training Loss > Validation Loss?
• SVM Kernel/Polynomial Trick
• Bias Variance Tradeoff
• Multi-GPU Parallelism

Speech

• Speech Processing

Vision

• Receptive Field
• Skip Connections
• Vision Transformer (ViT)

NLP

• Attention
• Transformers
• Autoregressive vs. Autoencoder Models
• BERT
• GPT
• BigBird
• Named Entity Recognition
• Token Sampling Methods: Greedy, Beam Search, Temperature, top-k, top-p
• Convolutional Neural Networks for Text Classification
• Relationship between Hidden Markov Model and Naive Bayes
• Maximum Entropy Markov Models and Logistic Regression
• Conditional Random Fields for Sequence Prediction

Multimodal

• Vision Language Models

Privacy-Preserving/On-Device AI

• Model Compression using Training/Inference Optimizations
• Personally Identifiable Information (PII)
• Federated Learning
• Differential Privacy

Evaluation

• Evaluation Metrics, ROC-Curves and Imbalanced Datasets
• F-Beta Score

MLOps/Production

• Concept/Data/Label Drift

Miscellaneous

• Chain Rule
• Bayes’ Theorem
• Probability Calibration
• N-Dimensional Tensor Product
• PyTorch vs. TensorFlow
• Similarity Search
• Transferability Estimation
• TensorBoard

Hyperparameters

• Hyperparameter Tuning
• Hyperparameter Logging

Practice

• Interview Questions