Please enable JavaScript.

Coggle requires JavaScript to display documents.

Artificial intelligence, AI (Projects (games (examples (deepmind…

- - - - Artificial intelligence refers to "a broad set of methods, algorithms and technologies that make software 'smart' in a way that may seem human-like to an outside observer," said Lynne Parker
    - - Acting Humanly (ponašati se ljudski)
        
        the turing test approach part of RA
        
        What?
        
        total turing test
        
        capabilities
        
        compoter vision
        
        robotics
        
        history
        
        Alan Turing ( 1950 )
        
        was designed to provide a satisfactory
        operational definition of intelligence.
        
        capabilities
        
        natural language processing
        
        knowladge representation
        
        automated reasoning
        
        machine learning
      - Thinking humanly (razmišljati ljudski)
        
        the cognitive modeling approach
        
        brings together
        
        psychology =>
        
        theory of human mind (cognitive science)
        
        computer model from AI
      - Thinking rationally (razmišljati racionalno)
        
        the "laws of thought" approach part of RA
        
        brings together => intelligent system
        
        logic
        
        AI
        
        history
        
        Aristotel => "right thinking"
      - Acting rationally (ponašati se racionalno)
        
        the rational agent(RA) approach
        
        limitet rationality
  - - - theories
        
        utility
        
        Desision theory
        
        combines probability theory with utility theory
        
        Game theory
    - - neuron
    - - cognitive science
    - - control theory
        
        history
        
        Norbert Wiener
        
        Ashby’s Design for a Brain (1948, 1952)
        
        elaborated on his idea that intelligence could be created by the use of homeostatic devices containing appropriate feedback loops to achieve stable adaptive behavior
    - - grew up hybrid field
        
        computational linguistics ( natural language proseccing )
- - - - definition
        
        "This is a subset of A.I. that deals with cognitive behaviors we associate with 'thinking' as opposed to perception and motor control," Dietterich
- - - - deepmind
        
        PySc2 for StarCraft II
        
        StarCraft AI
        
        agent
        
        what
        
        an intelligent program that acts autonomously in an enviroment
        
        solving several difficult problems
        
        planning
        
        optimization
        
        multiagent control
  - - - Evolution Strategies, ES
        
        scalable alternative to reinforcement learning(RL)
        
        optimizatio technique
  - - - ML from disaster => kaggle data project
- - - - preparing the data
      - training a model
      - evaluating the model
      - improving the performance
      - collecting data
    - - covering multiple technogies
    - - lerning methods
        
        supervised
        
        Data is labeled and comes in pairs (input, output)=(x, y). We need to find a mapping f(x) = y
        
        if y is a discrete label(nebrojčana vrijednost)
        
        classification
        
        If y is a number
        
        regression
        
        unsupervised
        
        Data comes without labels. We need to find regularities in the data
        
        density estimation :question:
        
        clustering
        
        novelty/outlier detection
        
        dimensionality reduction
        
        reinforcement learning
        
        Learning an optimal strategy based on trials with a delayed credit
      - how lear model?
        
        learning steps
        
        train
        
        About
        
        feature extractor (Ekstraktor značajki)
        
        1 more item...
        
        during training the model is fed with data
        
        1 more item...
        
        problem
        
        overfitting :red_cross:
        
        2 more items...
        
        test
        
        We train the model on the train set, use this model to obtain predictions on the test set, and then compute the accuracy
        
        the accuracy
        
        1 more item...
        
        create model
        
        prepare dataset for learning
        
        cross-validation (unakrsna validacija)
        
        split dataset
        
        2 more items...
        
        prediction
        
        during prediction, the model is presented only data instance x as input ⇒ generates y as output
    - - numerical cumputation
    - - Na ̈ıve Bayes Classifier
      - decision trees
  - - - What?
        
        that involves the grouping of data points
- - - - What?
        
        DeepMind Tehnologies
      - deep Q learning or deep reinforcement learning
        
        combination of deep learning and reinforcement learning
    - - What?
        
        combination of NN and fuzzy logic
    - - What?
        
        includes a large system of neurons arranged in several hidden layers
      - learning
        
        deep learning
    - - Supervised
        
        Adeline
        
        Percepton
        
        percepton rule ( pravilo perceptona)
        
        ukoliko se uzorak klasificira ispravno, ne radi korekciju.
        
        ukoliko se uzorak klasificira neispravno, primjeni LMS pravilo
        
        Ciklički uzimaj sve uzorke redom, a postupak zaustavi kada sve uzorke klasificiraš ispravno za redom
        
        temelji na matematičkom modelu McCulloch-Pittsa, poboljšao ga je Frank Rosenblatt
        
        Multilayer percepton, MLP (Višeslojni percepton )
        
        sastavljena od percepton čvorova ( nodes )
        
        raspoređenih u
        
        više slojeva
        
        izlazni
        
        ulazni
        
        skriveni
        
        dva sloja
        
        karakteristike
        
        gl. karakteristika
        
        potpuna povezanost njegovih čvorova
        
        znači da svaki čvor u svakom sloju povezan sa svim čvorovima predhodnog sloja
        
        signali
        
        vrste
        
        funkcijski signal ili ulazni signal
        
        1 more item...
        
        izlasni signal ( Signal greške )
        
        1 more item...
        
        signali kroz mrežu kreću s lijeva na desno ( unaprijed )
        
        učenje
        
        backpropaganation algorithm
    - - used for
        
        effective mechnism used for image recognition
      - idea
        
        inspired by brain
        
        visual cortex
        
        visual neuron cells in the brain that each act in a different way
        
        simple cells (S cells)
        
        What?
        
        1 more item...
        
        complex cells ( C cells)
        
        continue to respond to a certain stimulus, even though its absolute position on the retina changes
      - model
        
        hierarchical NN model
        
        neocognitron
      - history
        
        first
        
        LeNet-5
      - arhitecture
        
        layers
        
        organised in 3 dimensions
        
        height - H
        
        dept - D
        
        width -W
        
        components
        
        1) the hidden layer/ Feature extraction part
        
        2) the classification part
  - - - Umjetna neuronska mreža (engl. artificial neural networks, ANN) čini cjelinu međusobno povezanih čvorova čije veze možemo prilagođavati prema skupu podataka za učenje.Paradigma temeljena na biološkoj neuronskoj mreži ili mozgu znatno je pojednostavljena jer se nemogu programirati fenomeni koji posjeduje živčani sustav. Moć obrade podataka nalazi se u vezama između pojedinih procesnih jedinica.
    - - problem klasifikacija
      - predviđanje
      - vezane uz ulazne i izlazne varijable i probleme optimizacije
      - raspoznavanje uzoraka, obrada slika, govora, simulacija
      - obrada podataka nepreciznih i nekompletnih podataka
    - - Node, Process unit (Umjetni neuron, procesna jedinica ili čvor)
        
        sastoji
        
        Weight (težina)
        
        analogne dendritima
        
        težinske sume ili net
        
        analogna je somi ili tijelu živčane stanice
        
        net = w0x0 + w1x1 + w2x2 + ... + wnxn
        
        funkcije praga ili aktivacijske funkcije
        
        Treshold ( prag ) ili pomak (offset ili bias )
        
        uzimamo θ = - w0
        
        tipovi :
        
        step func.
        
        daje boolov iskaz
        
        1 more item...
        
        funkcija praga
        
        najčešći prijenos funkcije, daje neku prednost kod učenja NN
        
        f(net ) =
        
        sigmoidalna
        
        f(net) = {
        
        3 more items...
        
        analogna je integracijskom dijelu neurona
        
        varijable
        
        ulazna ( Signali ) ili funkcijski signal
        
        analogija
        
        sinapse kod neurona
        
        about
        
        su realni brojevi u intervalu [-1,1] ili [0,1]
        
        označavanje
        
        x1, x2, xn
        
        izlazna
        
        analogija
        
        s aksom
        
        označava s y
    - - broj slojeva procesnih jedinica
        
        u ANN procesne jedinice ( node) smještene su u slojeve
        
        dvoslojne
        
        sastoje se od dva sloja
        
        ulazni sloj
        
        izlazni sloj
        
        višeslojne
        
        sastoje se
        
        skriveni slojevi
        
        može ih biti više
        
        izlazni sloj
        
        ulazni sloj
      - prijenosne funkcije
      - tip veza između ulaznih i izlaznih podataka
        
        procesne jedinice u NN mogu biti spojene
        
        inter-slojna veza
        
        za povezivanje čvorova u razl. slojevima NN
        
        intra-slojna veza
        
        služe za povezivanje čvorova unutar jednog sloja
        
        primjer Hopfildova mreža
      - tip učenja
        
        kriterij određuje tip učenja => da li je izlazna vrijednost poznata ili ne ?
        
        iz nije poznata
        
        Nenadgledano (Unsupervised )
        
        suprotno nadgledanom učenju => sustav sam otkriva odnose između podataka prema opisanim značajkama
        
        koriste se
        
        2 more items...
        
        iz poznata
        
        Supervised (nadgledano )
        
        tip učenja kod kojeg sustav sam podešava parametre na temelju podataka s poznatim ulaznim i izlaznim vrijednostima
        
        koliko primjera za učenje pokažemo meži za vrijeme trajanja jedne iteracije
        
        Bactch ili off-line (grupno )
        
        predočavamo sve primjere u jednoj iteraciji, zbog toga NN uči samo u jednoj fazi u kojoj se prilagode težine, a u ostalim fazama težine su fiksirane
        
        Patern-By-Patern or on-line (pojedinačno)
        
        NN uči u vremenskim intervalima, u svakom vremenskom intervalu, tj. iteraciji dodjeljen je jedan primjer za učenje NN pritom se vrše prilagodbe težinskih faktora
      - prijenos informacija kroz NN
        
        FeedForward ( Statički )
        
        informacije su plasirane u više slojeve u jednom prolazu i najvažnije informacija se ne vraća u niže slojeve
        
        FeedBack ( Dinamičke )
        
        informacije se vraćaju u niže slojeve,
        
        informacije se mogu plasirati u dva smjera
        
        ulazne informacije primaju u vremenskim intervalima
    - - faza učenja ili treniranja
        
        neuronsku mrežu nemožemo učiti ako nismo definirali
        
        tip NN
        
        model ( ulazno - izlazne varijable )
        
        karakteristike
        
        Kako dugo trenirati neuronsku mrežu ?
      - faza testiranja
        
        proces ocjenjivanja mreže
        
        u ovoj fazi težine su fiksirane, te ocjenjujemo kako će se mreža ponašati s novim podacima
      - podjela prikupljenih podataka u podskupove #
- - - - computer vision (rač. vid )
      - patern repognition (prepoznavanje uzorka)
      - scene analysis (analiza scena)
    - - natural language procesing ,NLP
        
        components
        
        NL understanding, NLU
        
        NL generation, NLG
        
        involves
        
        text planing
        
        sentence planning
        
        text realization
    - - About
        
        state space search (pretraživanje prostora stanja)
        
        mathematic model
        
        search problem
        
        consists :
        
        3 more items...
        
        S - set of states (state space)
        
        methods
        
        search tree
        
        By searching through a digraph, we gradually construct a search tree
        
        1 more item...
        
        tree vs state space
        
        3 more items...
        
        (basic idea of state space search)
        
        directed graph or digraph (usmjereni graf)
        
        3 more items...
        
        search startegies
        
        blind or uninformed search (cro. slijepo pretraživanje)
        
        type
        
        1 more item...
        
        Heuristic (directed or informed) search (cro. usmjereno pretraživanje)
        
        type
        
        3 more items...
        
        control methods :question:
        
        problems
        
        puzzle
        
        comparing problems vs. algorithms
        
        Problem properties:
        
        Algorithm properties:
    - - chatbots
        
        About
        
        an AI software that can simulate a conversation (or chat) with a user in natural language through messaging applications, websites, mobile apps or through the telephone
        
        there are two different tasks at the core of a chatbot
        
        user request analysis
        
        returning the response
        
        2019 ...
        
        Watson Assistant
        
        Bold360
        
        Rulai
        
        LivePerson
        
        Inbenta
        
        Ada
        
        Vergic
    - - cognitive modeling
      - philosophical foundation
- - - - completely automated public turing test to tell computes and humans apart
  - - - Alan Tuning
        
        Can machines think?
        
        experiment
        
        The imitation game
- - - - nastaje na pojmu Fuzzy skupova (Fuzzy sets)
        
        klasični skupovi vs fuzzy skupova
        
        fuzzy skupovi
        
        nisu jasno određene granice
        
        Membership function ( funkcije pripadnosti )
        
        2 more items...
        
        klasični skupovi
        
        jasno određene granice
        
        sastoje od elemenata koji po nekom kriteriju pripadaju u taj skup
      - Fuzzy interence system, FIS (Fuzzy sustav za zaključivanje)
        
        model
        
        input
        
        output
        
        fuzzy baza podataka
        
        povezana sa :
        
        dio za zaključivanje # ( Agregacija)
        
        više modela :
        
        2 more items...
        
        about
        
        1 more item...
        
        Agregacija
        
        2 more items...
        
        defazifikacija
        
        suprotan fazifikaciji,
        
        1 more item...
        
        dobiva se koeficijent
        
        1 more item...
        
        fazifikacija
        
        prvi korak, tansformiranje crisp ulaznih parametara u fuzzy ulazne parametre
        
        1 more item...
        
        je sustav za računanje koji raspolaže s više koncepata znanja, a to su :
        
        teorija fuzzy skupova
        
        fuzzy if-then pravila
        
        lingvistička varijabla
        
        " riječ " subjektivnog karaktera
        
        if -then pravila => pravila po kojima fuzzy logika donosi zaključke
        
        jezičnog karaktera
        
        sastoje :
        
        premisa
        
        zaključaka
        
        fuzzy zaključivanje
- - - - matrices
        
        WHAT ?
        
        2D array of numbers, , so each element is identified by two indices instead of just one
        
        an example of 2nd-order tensors
      - vectors
        
        About
        
        what?
        
        is an array of numbers, numbers are arranged in order
        
        an example of 1st-order tensor
      - tensors (cro. tenzor)
        
        What?
        
        In some cases we will need an array with more than two axes
      - scalars
        
        About
        
        What?
        
        is just a single number
        
        an example of a 0th-order tensor
        
        built-in scalar types
        
        int
        
        float
        
        complex
        
        bytes
        
        Unicode ( in Python )
  - - - uncertainty
        
        possible source of uncertainety
        
        2.
        
        Incomplete observability
        
        3.
        
        Incomplete modeling
        
        1.
        
        Inherent stochasticity in the system being modeled
      - softmax function
        
        often used to predict the probabilities associated with a multinoulli distribution
    - - poor conditioning
      - Gradient-Based Optimization
- - - - cross-entropy
      - decision trees on basic of maximum information gain
      - Viterbi algorithm
        
        widely used in NLP and Speech
      - machine tlanslatation => ENCODER - DECODER