Machine Learning Hub
Machine Learning (ML) is the field of study that gives computers the ability to learn without being explicitly programmed. This hub organizes Wikantik's ML content, providing a path from mathematical foundations through modern deep learning architectures to the practical realities of deploying models in production.
Foundations
The conceptual and mathematical bedrock of learning from data.
- [Machine Learning](MachineLearning) — Comprehensive technical overview of ML paradigms and foundational principles
- [Deep Learning Fundamentals](DeepLearningFundamentals) — Conceptual foundations of neural networks, backpropagation, and the loss landscape
- [Mathematical Foundations of Machine Learning](MathematicalFoundationsOfMachineLearning) — The linear algebra, calculus, and probability that make learning possible
- [Regularization Techniques](RegularizationTechniques) — Methods to prevent overfitting and ensure models generalize to new data
Architectures
The structural blueprints for modern neural networks.
- [Neural Network Architectures](NeuralNetworkArchitectures) — Survey of MLPs, CNNs, RNNs, and the dominant Transformer architecture
- [Transformer Architecture](TransformerArchitecture) — Deep dive into the self-attention mechanism that revolutionized NLP and vision
- [Convolutional Neural Networks](ConvolutionalNeuralNetworks) — Architectures designed for spatial data like images
- [Recurrent Neural Networks](RecurrentNeuralNetworks) — Models for sequential data and time series
Core Techniques and Training
The engineering and optimization processes used to build effective models.
- [Gradient Descent and Optimizers](GradientDescentAndOptimizers) — How models learn by traversing the loss landscape
- [Feature Engineering](FeatureEngineering) — Transforming raw data into informative signals for the model
- [Model Selection](ModelSelection) — Strategies for picking the right model class for a given problem
- [Cross-Validation and Model Evaluation](CrossValidationAndModelEvaluation) — Rigorous methods for measuring how well a model will perform in the real world
- [Model Selection Efficiency](ModelSelectionEfficiency) — Techniques for searching the hyperparameter space effectively
Specialized Domains
Applying machine learning to specific types of data and problems.
- [Natural Language Processing](NaturalLanguageProcessing) — Modeling human language: from embeddings to Large Language Models (LLMs)
- [Computer Vision Fundamentals](ComputerVisionFundamentals) — Processing visual data: classification, detection, and segmentation
- [Reinforcement Learning Fundamentals](ReinforcementLearningFundamentals) — Learning through interaction with an environment to maximize rewards
- [Recommendation Systems](RecommendationSystems) — Collaborative filtering, content-based filtering, and ranking at scale
- [Sentiment Analysis with Machine Learning](SentimentAnalysisWithMachineLearning) — Extracting emotional tone from text
- [Anomaly Detection Techniques](AnomalyDetectionTechniques) — Identifying outliers and suspicious patterns in datasets
- [TimeSeries Forecasting](TimeSeriesForecasting) — Predicting future values in sequential data
- [Text Analysis with Data Science](TextAnalysisWithDataScience) — Extracting structured insights from unstructured text
MLOps and Deployment
The practical discipline of serving and maintaining models in production environments.
- [ML Model Deployment](MlModelDeployment) — Packaging, versioning, and rolling out models to production
- [ML Model Deployment Hub](MlModelDeploymentHub) — Specialized index for model serving and lifecycle management
- [MLOps Practices](MLOpsPractices) — The intersection of ML, DevOps, and Data Engineering
- [Inference Serving](InferenceServing) — Infrastructure and patterns for low-latency model predictions
- [Cost Effective Inference](CostEffectiveInference) — Strategies for reducing the compute cost of running models
- [CPU Inference](CPUInference) — Optimizing model execution on traditional CPU hardware
- [GPU Acceleration](GPUAcceleration) — Leveraging specialized hardware for training and high-throughput inference
Advanced and Emerging Topics
Pushing the boundaries of model capability and efficiency.
- [Fine-Tuning Large Language Models](FineTuningLargeLanguageModels) — Adapting massive pretrained models to specific downstream tasks
- [Model Quantization](ModelQuantization) — Compressing models for efficient edge and mobile deployment
- [Synthetic Data Generation](SyntheticDataGeneration) — Creating artificial training data to supplement real-world datasets
- [The Future of Machine Learning](TheFutureOfMachineLearning) — Emerging trends and the research trajectory of the field
Adjacent Hubs
- [Mathematics Hub](MathematicsHub) — The formal language and proofs underlying ML theory
- [Data Engineering Hub](DataEngineeringHub) — Building the pipelines that feed models with high-quality data
- [Java Hub](JavaHub) — Implementing ML services and backends on the JVM