L5 Flashcards
def NLP
branch of CS (AI) that gives computers ability to understand text and spken words and respond with some of their own
- tries to brake it into shorter pieces to understand relationship between them
- mix of NLU and NLG
Challenges of NLP
- human language is filled with ambiguities
- irregularities: homonyms, homophones, scarcasm, idios, metapgors, exceptions, variation in sentence structure
4 stages of NLP
- morphological processing: break strings into tokens (smaller pieces)
- syntax analysis: check for and recognize grammar/sentence structure
- semantic analysis: meaning
- pragmatic analysis: applied context
7 stages of NLP test preprocessing
watch yt video to complete
- segmenting: break into sentences
- tokenizing: break into words
- stop word: prepositions
- stemming: same words with different prefit or suffix
- lemmatization: multiple words can have same meaning
- speech tagging: words are tagges as nouns, verb, preposition, etc
- named entity tagging: identify some words that may occur in documents
NLU
natural language understanding: receives and interprets written/spoken input
NLG
natural language generation: system responds to message it receives by speaking or writting answer
NLP use cases (6)
- spam detection
- translation
- virual agent and chatbots
- searching engine
- social media sentiment analysis
- text extraction
Computer vision
what is it
examples
- interdisciplinary scientific field that works on computers gaining high-level understanding from digital images or videos
- understand and automate tasks that the human visual system can do
- includes methods for acquiring, processing , analyzing and understanding digital images and extracting high-dimensional data from it producing numerical or symbolic information
eg) image classification, object/keypoint detection, image segmentation, tracking
Computer vision approaches
ML: with enough data models can learn to distinguish images from each other
DL: CLL and RNN
CNN
convolutional neural networks
- recognizes images
- first discerns hard edges and simple shapes and then fills in info in more iterations
RNN
recurrent neural network
- recognized videos
- same but with single frames from videos
Computer vision applications
- transport: self-driving cars
- healthcare: x-ray, CT and MRI scans
- manufacturing: defect product inspection, text and barcodes
- construction: predictive maintenance
- agriculture: crop and yield monitoring, insect detection, disease detection
- retail: self-checkout, smart store
Robotics
- interdisciplinary sector of science and engineering that deals with design, construction, operation and use of robots for control, sensory feedback and information processing
- replace human where they don’t want to or can’t be
Types of Robotics (5)
- Preprogrammed robots
- Humanoid robots
- Autonomous robots
- Teleoperated robots
- Augmenting robots
Pre-programmed robots
- robot arms in the automotive assembly line
- automated guided vehicles (AGV)
- cobots
Humanoid robots
- look like or mimic human behaviour
- han robotics sophia
- boston dynamics’ atlas
Autonomous Robots
- operate independently of human operators
- designed to carry out tasks in open environments
- use of sensors
eg ) cleaning bots, autonomous drones
Teleoperated robots
robots control over a wireless network from safe distance
Augmenting robot
enhance or replace capabilities of humans
eg) robotic legs
Industry Applications Robots (6)
- Manufacturing: deployed to expedite processes, drive efficiency and promote safety
- Farming and agriculture: help famers havest their crops more quickly and efficiently, assess ripeness, move branches, pick up food
- Logistics: help shipping and delivery to be quick and efficient
- Healthcare: surgery, medication delivery
- Retail, customer service
- Smart cities: way-finding and information services, security patrols, building construction
- Home, entertainment
- Onlien robots: chatbots
