[{"data":1,"prerenderedAt":608},["ShallowReactive",2],{"$O-HKhpFFdB":3},{"id":4,"title":5,"body":6,"date":597,"description":20,"extension":598,"meta":599,"navigation":603,"path":604,"seo":605,"stem":606,"__hash__":607},"articles\u002Farticles\u002F50-ai-concepts.md","50 AI Essential Concepts — Quick Reference",{"type":7,"value":8,"toc":576},"minimark",[9,14,21,24,29,36,38,42,104,106,110,132,134,138,158,163,220,222,226,264,266,270,296,298,302,334,336,340,366,368,372,426,428,432,452,454,458,478,480,484,516,518,522,542,544,548,568,570],[10,11,13],"h1",{"id":12},"_50-ai-essential-concepts-quick-reference-summary","🤖 50 AI Essential Concepts — Quick Reference Summary",[15,16,17],"blockquote",{},[18,19,20],"p",{},"A concise cheat sheet covering the core ideas in modern AI\u002FML. Save this for quick revision.",[22,23],"hr",{},[25,26,28],"h2",{"id":27},"what-is-ai","🧠 What is AI?",[18,30,31,35],{},[32,33,34],"strong",{},"Artificial Intelligence"," replicates human cognitive functions — reasoning, learning, problem-solving, and decision-making — ranging from simple rule-based systems to complex adaptive neural networks.",[22,37],{},[25,39,41],{"id":40},"core-learning-paradigms","📚 Core Learning Paradigms",[43,44,45,61],"table",{},[46,47,48],"thead",{},[49,50,51,55,58],"tr",{},[52,53,54],"th",{},"Type",[52,56,57],{},"How it works",[52,59,60],{},"Use case",[62,63,64,78,91],"tbody",{},[49,65,66,72,75],{},[67,68,69],"td",{},[32,70,71],{},"Supervised Learning",[67,73,74],{},"Learns from labeled input\u002Foutput pairs",[67,76,77],{},"Image classification, spam detection, price prediction",[49,79,80,85,88],{},[67,81,82],{},[32,83,84],{},"Unsupervised Learning",[67,86,87],{},"Finds hidden patterns in unlabeled data",[67,89,90],{},"Clustering, anomaly detection, dimensionality reduction",[49,92,93,98,101],{},[67,94,95],{},[32,96,97],{},"Reinforcement Learning",[67,99,100],{},"Agent learns by receiving rewards\u002Fpenalties from actions",[67,102,103],{},"Game playing, robotics, adaptive systems",[22,105],{},[25,107,109],{"id":108},"machine-learning-foundations","🔬 Machine Learning Foundations",[111,112,113,120,126],"ul",{},[114,115,116,119],"li",{},[32,117,118],{},"Machine Learning"," — Systems improve performance through experience; backbone of modern AI.",[114,121,122,125],{},[32,123,124],{},"Pattern Recognition"," — Identifies trends and relationships in large datasets.",[114,127,128,131],{},[32,129,130],{},"Predictive Analytics"," — Uses historical data to make informed future decisions.",[22,133],{},[25,135,137],{"id":136},"neural-networks","🧬 Neural Networks",[111,139,140,146,152],{},[114,141,142,145],{},[32,143,144],{},"Neural Networks"," — Inspired by the brain; artificial neurons receive inputs, apply weights, and use activation functions to produce outputs.",[114,147,148,151],{},[32,149,150],{},"Architecture layers",": Input layer → Hidden layers (with activation functions) → Output layer.",[114,153,154,157],{},[32,155,156],{},"Deep Learning"," — Multiple hidden layers extract abstract features; enables breakthroughs in image\u002Fspeech\u002Flanguage.",[159,160,162],"h3",{"id":161},"key-network-types","Key Network Types",[43,164,165,175],{},[46,166,167],{},[49,168,169,172],{},[52,170,171],{},"Network",[52,173,174],{},"Speciality",[62,176,177,188,199,210],{},[49,178,179,185],{},[67,180,181,184],{},[32,182,183],{},"CNN"," (Convolutional)",[67,186,187],{},"Vision & pattern recognition (facial recognition, medical imaging, self-driving)",[49,189,190,196],{},[67,191,192,195],{},[32,193,194],{},"RNN"," (Recurrent)",[67,197,198],{},"Sequential\u002Ftime-series data (text, audio, video)",[49,200,201,207],{},[67,202,203,206],{},[32,204,205],{},"LSTM"," (Long Short-Term Memory)",[67,208,209],{},"Long-range dependencies; solves RNN's vanishing gradient problem",[49,211,212,217],{},[67,213,214],{},[32,215,216],{},"Transformer",[67,218,219],{},"Language processing via self-attention; parallel processing; powers LLMs",[22,221],{},[25,223,225],{"id":224},"training-optimization","🔧 Training & Optimization",[111,227,228,234,240,246,252,258],{},[114,229,230,233],{},[32,231,232],{},"Gradient Descent"," — Iteratively adjusts parameters to minimize error\u002Floss.",[114,235,236,239],{},[32,237,238],{},"Backpropagation"," — Propagates error backward through layers to update weights.",[114,241,242,245],{},[32,243,244],{},"Hyperparameters"," — Settings that shape training: learning rate, architecture, batch size, epochs.",[114,247,248,251],{},[32,249,250],{},"Overfitting"," — Model memorizes training data; fix with regularization, cross-validation, more data.",[114,253,254,257],{},[32,255,256],{},"Underfitting"," — Model too simple; fix by increasing complexity or training longer.",[114,259,260,263],{},[32,261,262],{},"Cross-Validation"," — Splits data into folds to get a reliable performance estimate.",[22,265],{},[25,267,269],{"id":268},"️-data-features","🛠️ Data & Features",[111,271,272,278,284,290],{},[114,273,274,277],{},[32,275,276],{},"Training Data"," — Quality > quantity; needs clean, diverse, balanced, labeled examples.",[114,279,280,283],{},[32,281,282],{},"Feature Engineering"," — Transform raw data into meaningful representations; select the most relevant variables.",[114,285,286,289],{},[32,287,288],{},"Big Data"," — Petabyte-scale, high-velocity, varied formats (structured\u002Funstructured\u002Fsemi-structured).",[114,291,292,295],{},[32,293,294],{},"Bias in AI"," — Skewed training data or design choices can cause unfair treatment of groups.",[22,297],{},[25,299,301],{"id":300},"advanced-techniques","🚀 Advanced Techniques",[111,303,304,310,316,322,328],{},[114,305,306,309],{},[32,307,308],{},"Transfer Learning"," — Reuse a pre-trained model's knowledge for a new task; reduces data & time needs.",[114,311,312,315],{},[32,313,314],{},"Fine-Tuning"," — Freeze early layers, retrain final layers for domain-specific accuracy.",[114,317,318,321],{},[32,319,320],{},"Ensemble Methods"," — Combine multiple models (voting\u002Faveraging) to reduce errors and boost robustness.",[114,323,324,327],{},[32,325,326],{},"Random Forests"," — Ensemble of decision trees using bootstrap sampling + majority vote.",[114,329,330,333],{},[32,331,332],{},"Attention Mechanisms"," — Dynamically assign weights to input elements based on relevance; key to transformers.",[22,335],{},[25,337,339],{"id":338},"️-language-vision","🗣️ Language & Vision",[111,341,342,348,354,360],{},[114,343,344,347],{},[32,345,346],{},"NLP (Natural Language Processing)"," — Extracts meaning, sentiment, and intent from text; powers translation and text generation.",[114,349,350,353],{},[32,351,352],{},"Large Language Models (LLMs)"," — Massive models (e.g., GPT-3: 175B parameters) trained on huge text corpora; multilingual and generative.",[114,355,356,359],{},[32,357,358],{},"Generative AI"," — Creates new text, images, audio, and video by learning patterns from large datasets.",[114,361,362,365],{},[32,363,364],{},"Computer Vision"," — Interprets visual data; enables object detection, facial analysis, medical diagnosis, and autonomous navigation.",[22,367],{},[25,369,371],{"id":370},"classic-ai-approaches","🤖 Classic AI Approaches",[43,373,374,384],{},[46,375,376],{},[49,377,378,381],{},[52,379,380],{},"Approach",[52,382,383],{},"Description",[62,385,386,396,406,416],{},[49,387,388,393],{},[67,389,390],{},[32,391,392],{},"Expert Systems",[67,394,395],{},"Knowledge base + inference engine; solves specialized problems using human-encoded rules",[49,397,398,403],{},[67,399,400],{},[32,401,402],{},"Fuzzy Logic",[67,404,405],{},"Handles degrees of truth (0–1); used in control systems and risk assessment",[49,407,408,413],{},[67,409,410],{},[32,411,412],{},"Genetic Algorithms",[67,414,415],{},"Mimics natural evolution (selection, crossover, mutation) to optimize solutions",[49,417,418,423],{},[67,419,420],{},[32,421,422],{},"Swarm Intelligence",[67,424,425],{},"Emergent behavior from simple agents (ant colony, particle swarm, bee algorithms)",[22,427],{},[25,429,431],{"id":430},"classical-ml-algorithms","📐 Classical ML Algorithms",[111,433,434,440,446],{},[114,435,436,439],{},[32,437,438],{},"SVM (Support Vector Machine)"," — Finds the optimal hyperplane maximizing class margin; uses kernel trick for non-linear data.",[114,441,442,445],{},[32,443,444],{},"K-Means Clustering"," — Assigns points to k centroids iteratively until clusters stabilize.",[114,447,448,451],{},[32,449,450],{},"PCA (Principal Component Analysis)"," — Reduces data dimensions while preserving maximum variance.",[22,453],{},[25,455,457],{"id":456},"ai-deployment-infrastructure","🌐 AI Deployment & Infrastructure",[111,459,460,466,472],{},[114,461,462,465],{},[32,463,464],{},"Edge Computing"," — Processes data locally on-device; near-instant response, privacy-preserving, low bandwidth.",[114,467,468,471],{},[32,469,470],{},"Cloud AI"," — Scalable GPU clusters; elastic scaling; global access via API endpoints.",[114,473,474,477],{},[32,475,476],{},"Federated Learning"," — Devices train locally, share model updates (not raw data); preserves privacy.",[22,479],{},[25,481,483],{"id":482},"specialized-ai-types","🧩 Specialized AI Types",[111,485,486,492,498,504,510],{},[114,487,488,491],{},[32,489,490],{},"Narrow AI"," — Excels at one specific task (chess, voice assistants, recommendations); no general adaptability.",[114,493,494,497],{},[32,495,496],{},"AGI (Artificial General Intelligence)"," — Theoretical human-level reasoning across all domains; not yet achieved.",[114,499,500,503],{},[32,501,502],{},"Robotics"," — AI + sensors + actuators for autonomous physical operation.",[114,505,506,509],{},[32,507,508],{},"Autonomous Systems"," — Self-operating systems in transport, logistics, agriculture, and exploration.",[114,511,512,515],{},[32,513,514],{},"Multimodal AI"," — Processes text, images, audio, and video simultaneously.",[22,517],{},[25,519,521],{"id":520},"️-ethics-trust","⚖️ Ethics & Trust",[111,523,524,530,536],{},[114,525,526,529],{},[32,527,528],{},"Explainable AI (XAI)"," — Makes model decisions transparent (LIME, SHAP, attention visualization).",[114,531,532,535],{},[32,533,534],{},"AI Ethics"," — Fairness, privacy, accountability, and transparency; aligning AI with human values.",[114,537,538,541],{},[32,539,540],{},"Bias"," — Data bias (skewed datasets) and algorithmic bias (design amplifying certain outcomes) require careful mitigation.",[22,543],{},[25,545,547],{"id":546},"emerging-trends","🔮 Emerging Trends",[111,549,550,556,562],{},[114,551,552,555],{},[32,553,554],{},"Neuromorphic Computing"," — Chips that mimic brain structures for energy efficiency.",[114,557,558,561],{},[32,559,560],{},"Quantum Machine Learning"," — Potential exponential speed-ups using quantum computation.",[114,563,564,567],{},[32,565,566],{},"AI–Human Collaboration"," — Augmenting human capabilities, not replacing them.",[22,569],{},[18,571,572],{},[573,574,575],"em",{},"Source: \"50 AI Essential Concepts\" video captions summary.",{"title":577,"searchDepth":578,"depth":578,"links":579},"",2,[580,581,582,583,587,588,589,590,591,592,593,594,595,596],{"id":27,"depth":578,"text":28},{"id":40,"depth":578,"text":41},{"id":108,"depth":578,"text":109},{"id":136,"depth":578,"text":137,"children":584},[585],{"id":161,"depth":586,"text":162},3,{"id":224,"depth":578,"text":225},{"id":268,"depth":578,"text":269},{"id":300,"depth":578,"text":301},{"id":338,"depth":578,"text":339},{"id":370,"depth":578,"text":371},{"id":430,"depth":578,"text":431},{"id":456,"depth":578,"text":457},{"id":482,"depth":578,"text":483},{"id":520,"depth":578,"text":521},{"id":546,"depth":578,"text":547},"2025-03-24","md",{"tags":600,"featured":603},[601,118,156,602,364],"AI","NLP",true,"\u002Farticles\u002F50-ai-concepts",{"title":5,"description":20},"articles\u002F50-ai-concepts","eGr_69Tif_6lInzKuXzZWn6Mrw9dF7F4pxBtYmYTnRQ",1776571073238]