Vista previa parcial del texto
¡Descarga La Inteligencia Artificial y más Esquemas y mapas conceptuales en PDF de Informática solo en Docsity!
What is machine learning? + Creating predictive models by finding relationships in data 1. Training data 2. Algorithm 3. Model (past observations) (Generalizes the relationship (encapsulates the function) Bun20), y between x and y as a function) Training xl, y A [x1X2:X3], y + a FE pt [x1X2:X31, y =— 1 Features Label | 10) 19) Dx] Y 4. Inferencing data Prediction (unlabeled features) — (inferred label) Inferencing Types of machine learning Machine Learning A 5 y] Supervised machine learning Unsupervised machine learning Training data includes known labels Training data ¡s unlabeled A, Clustering | Similar items are Regression Classification grouped together Label is a numeric value Label is a categorization (or class) = IA PY Y y Y a a Binary classification Muilticlass classification + E3 e el Label ís oris nota class Labelis one of multiple classes S Predict the number ofice O Oy S € Separate plants into groups creams sold based on »I Z based on common ya Sy) day, season, and weather És characteristics Deep learning Human neural network Artificial neural network << SF wo) + Each neuron is a function that operates on an input value (x) and a weight (w) + The function is wrapped in an activation function that determines whether to pass the output on » Neurons fire in response to electrochemical stimuli - When fired, the signal is passed to connected neurons Deep learning: neural network example - multiclass 4 «P(y=01x):02 + + P(y=1|x):07 e -P(y=21x):0.1 4 Xx 37.3 $ cxc168 "Xx: 19.2 + X4: 30.0 What is Azure Machine Learning? + Azure Machine Learning is a 5 z cloud-based platform for CODO machine learning. se a NN + Azure Machine Learning Studio 2 al SE is a user interface for accessing mas : Azure Machine Learning - ¡ : capabilities. i Y esmonoa —oomesro — ossma00s cosas + Machine learning models al == trained with Azure Machine has Learning can be published as Ez services. po | - | Detecting faces with the Face Service Everyone can use the Face service to detect: + Blur - Exposure - Glasses - Head pose - Noise - Occlusion Only Managed Microsoft customers can access facial recognition capabilities: - Similarity matching + Identity verification Reading text with Optical Character Recognition (OCR) + Detect the location and characters of printed and handwritten text Options for quick text extraction from images, or asynchronous analysis of larger scanned documents Shopping List Non-fat milk Bread MW hxure Al Studio. All hubs. Al Services — Vision + Document Extract text from images. OS O GPr-4 Vision O Phic3ision O Gro Detected attributes ¿SON Sample Image 3 Document intelignce 3 Nutrition facts lla AmountPerServing D recipes Servingsize Ybar(409) Serving erPackage:4 Totalfatiag Saturatedfat1 59 AmountPerServing Transfatog aloes19o Bus Monigage forms Cholesterlomg a cresroma no Marie cefcates id B Creditcards D Contras B Read B Layout mal valuesarebasedon VitaminAS0% VAN caloneciet Microsoft + Natural language processing concepts Learning Objectives + Natural language processing capabilities in Azure What is natural language processing (NLP)? a Text analysis Tokenization Opinion mining Training oo y led Machine Language model translation Conversational Al NLP and conversational Al in Azure b ta Speech Translator + Textto speech + Speech to text translation identification 
