Doc. Ing. Kristína Machová, PhD.

Symbolická UI
1. Turingov stroj a Turingov test

A Turing machine is a theoretical computing device that is able to read and write
symbols on a tape, and to move the tape left or right based on a set of rules. It is a
simple model of a computer that is used to understand the fundamental principles of
computation and to study the limits of what can be computed.

The Turing test is a method for evaluating the intelligence of a machine. It is based on
the idea that if a machine can engage in a conversation with a human in such a way that
the human cannot tell that they are talking to a machine, then the machine can be
considered intelligent.

2. Aké typy rozumných agentov poznáte

In artificial intelligence, an agent is a system that is able to perceive its environment and
take actions to achieve a goal

1. Simple reflex agents: These agents are based on simple reflexes and do not have
any internal state. They act based on the current state of the environment and do
not consider past experiences.
2. Model-based agents: These agents maintain a model of the environment and use
this model to make predictions and plan their actions.
3. Goal-based agents: These agents have a specific goal in mind and take actions
to try to achieve this goal.
4. Utility-based agents: These agents assign a value or utility to each possible
action and choose the action with the highest expected utility.
5. Learning agents: These agents are able to learn from their experiences and
improve their performance over time.
6. Autonomous agents: These agents are able to operate independently and make
decisions on their own, without direct human intervention.
7. Multi-agent systems: These systems consist of multiple agents that interact with
each other and the environment to achieve a common goal.
8. Bounded rational agents: These agents make decisions based on limited
information and computational resources, rather than considering all possible
options.
9. Rational agents: These agents make decisions based on a systematic and logical
process, considering all available information and options.
3. Kedy, kým a pri akej príležitosti bol uvedený pojem umelá inteligencia
The term "artificial intelligence" was coined in 1956 by John McCarthy, a computer
scientist at Dartmouth College, during a conference on the topic. McCarthy defined
artificial intelligence as "the science and engineering of making intelligent machines."

4. Definujte stavový priestor

A state space is a representation of the possible states that a system can be in, along
with the transitions between those states.

In a state space, the states of the system are represented by points, and the transitions
between states are represented by edges. The edges can be labeled with the conditions
or actions that cause the transition to occur.

For example, consider a simple system that has two states: "on" and "off". The state
space for this system would consist of two points, representing the two states, and one
edge connecting the two points. The edge could be labeled with the condition "flip
switch" to indicate that flipping the switch causes the system to transition from one
state to the other.

A state space can be used to represent a wide range of systems, including physical
systems, software systems, and biological systems. It is a useful tool for understanding
and predicting the behavior of complex systems, and it can be used to design control
systems that can regulate and optimize the performance of those systems.
5. Ako môže byť definovaný problém, ktorý riešime
A problem that we are trying to solve can be defined as a gap between the current state
and a desired state, and the goal of problem-solving is to find a solution that closes this
gap.

There are several key elements to consider when defining a problem that we are trying
to solve:

1. The current state: This is the starting point, or the situation we are in right now. It
is important to clearly define and understand the current state in order to identify
the problem and the gap between the current and desired states.
2. The desired state: This is the state that we want to reach or the goal we are trying
to achieve. It is important to define the desired state clearly and specifically in
order to have a clear target to work towards.
3. The gap: The gap is the difference between the current and desired states. It is
important to identify the gap in order to understand what needs to be done to
solve the problem.
4. The constraints: Constraints are any factors that limit our ability to solve the
problem or that must be taken into account when finding a solution. It is
important to identify and understand the constraints in order to find a feasible
solution.

6. Typy prehľadávania stavového priestoru do šírky

1. Breadth-First Search.It is a type of search algorithm that begins at the root node
and explores all the neighboring nodes at the present depth before moving on to
the nodes at the next depth level.
2. Uniform-cost search: This type of BFS expands the nodes in the state space in
order of increasing cost, meaning that it always chooses the cheapest node to
expand first.
3. Depth-limited search: This type of BFS is similar to iterative deepening search,
but it only searches to a fixed depth level and does not continue searching at
deeper levels if the goal is not found.
4. Bidirectional search: This type of BFS searches the state space from both the
start and the goal simultaneously, and it stops when the two searches meet in
the middle. It is useful for finding solutions in large state spaces, as it reduces
the search time by half.
7. Typy prehľadávania stavového priestoru do hĺbky

1. Depth-first search: This is the most basic form of DFS, in which the algorithm
explores the nodes of the state space in a depth-first manner, following a path to
its maximum depth before backtracking and exploring other paths.
2. Depth-first search with memoization: This type of DFS uses a technique called
memoization to store the results of subproblems in order to avoid re-computing
them. It can be used to improve the efficiency of DFS in problems with
overlapping subproblems.
3. Iterative deepening search: This type of DFS uses a depth-first search approach,
but it repeats the search at increasing depth levels until the goal is found. It is
useful for finding solutions in large state spaces, as it avoids the cost of
expanding all the nodes at once.
4. Depth-limited search: This type of DFS is similar to iterative deepening search,
but it only searches to a fixed depth level and does not continue searching at
deeper levels if the goal is not found.

8. Riešenie problému rozkladom

Problem decomposition is a technique for solving problems by breaking them down into
smaller pieces that can be solved independently, and then combining those solutions to
find a solution to the original problem.

There are several key considerations when using problem decomposition to solve a
problem:

1. Identifying subproblems: It is important to identify the subproblems that make up

the larger problem, and to ensure that each subproblem is well-defined and
independent from the others.
2. Choosing the right level of granularity: The subproblems should be small enough to
be manageable, but not so small that solving them does not make a significant
contribution to solving the overall problem.
3. Ordering the subproblems:It is important to choose an order that allows for the
most efficient solution.
4. Combining the solutions: Once the subproblems have been solved, it is necessary
to combine their solutions to find a solution to the original problem.
9. Heuristické prehľadávanie priestoru pojmov (hľadanie najskôr najlepší, lačné hľadanie)
Heuristic search is a type of search algorithm that uses heuristics, or "rules of thumb," to
guide the search for a solution to a problem. Heuristic search algorithms try to find a
solution that is "good enough" rather than an optimal solution, and they are often used in
situations where finding the optimal solution is computationally infeasible.

There are several different types of heuristic search algorithms, including:

1. Best-first search: Best-first search is a heuristic search algorithm that expands the
nodes in the state space in order of their estimated cost to the goal. It always
chooses the node that is estimated to be the closest to the goal, based on a
heuristic function that estimates the cost of reaching the goal from each node.
2. Greedy search: Greedy search is a heuristic search algorithm that always chooses
the node that appears to be the best option at the current moment, without
considering the long-term consequences of that choice. It may not always find the
optimal solution, but it is often faster than other heuristic search algorithms.

Heuristic search algorithms can be an effective tool for solving problems in artificial
intelligence and computer science, as they can find good solutions quickly in situations
where an optimal solution is not necessary or is not possible to compute.

10. Heuristické prehľadávanie priestoru pojmov (algoritmus A*)

The A* search algorithm uses a heuristic function, also known as a "heuristic estimate," to
estimate the cost of reaching the goal from each node. The heuristic function should be
designed to underestimate the actual cost of reaching the goal, in order to ensure that A*
search finds the optimal solution if one exists.
11. Heuristická funkcia (skutočná heuristická funkcia, odhad heuristickej funkcie a jej zložky)
A heuristic function, also known as a "heuristic estimate," is a function that is used to
estimate the cost of reaching the goal from a given node in a state space. It is used by
heuristic search algorithms, such as A* search, to guide the search for a solution to a
problem.

There are two main types of heuristic functions:

1. Actual heuristic function: The actual heuristic function is a function that is

designed to accurately estimate the cost of reaching the goal from a given node. It
is based on accurate and up-to-date information about the problem, and it is used
to guide the search for the optimal solution.
2. Estimated heuristic function: The estimated heuristic function is a function that is
designed to underestimate the actual cost of reaching the goal from a given node.
It is used to ensure that the A* search algorithm finds the optimal solution if one
exists.

A heuristic function is made up of several components, including:

1. The cost of reaching the node: This is the actual cost of reaching the node from
the start state, and it is typically calculated using a cost function that assigns a
cost to each action.
2. The heuristic estimate of the cost to the goal: This is the estimated cost of
reaching the goal from the node, based on a heuristic function that estimates the
cost of reaching the goal from each node.
3. The total estimated cost: The total estimated cost is the sum of the cost of
reaching the node and the heuristic estimate of the cost to the goal. It is used to
guide the search for a solution and to determine the order in which the nodes are
expanded.

12. Heuristiky zmenšujúce stavový priestor, orezaný stavový priestor

Heuristics that reduce the state space, also known as "pruning techniques," are
techniques that are used to reduce the size of the state space in a problem-solving
context. They are used to improve the efficiency of search algorithms by eliminating
unnecessary or irrelevant nodes from the state space, thus reducing the number of nodes
that need to be expanded.
Znalostné systémy
1. Definícia ZS a ich vlastnosti v porovnaní s ľudským riešiteľom
Knowledge-based systems, also known as expert systems, are artificial intelligence (AI)
systems that are designed to replicate the knowledge and expertise of a human expert in
a specific domain.

There are several key characteristics or properties of knowledge-based systems that

distinguish them from human experts:

1. Autonomy: Knowledge-based systems are able to operate independently, without

the need for direct human supervision or intervention.
2. Consistency: Knowledge-based systems are able to consistently apply the same
set of rules or algorithms to a problem, ensuring that they make the same decision
each time they encounter a similar situation.
3. Speed: Knowledge-based systems are able to process and analyze data much
faster than humans, allowing them to quickly arrive at a decision or solution.
2. Základná architektúra ZS a typy znalostí používaných v ZS
The basic architecture of a knowledge-based system is a knowledge base, an inference
engine, and a user interface.

1. Knowledge base contains all the information and knowledge that the system uses
to perform its tasks.
2. Inference engine: It applies a set of rules or algorithms to the knowledge base to
infer new conclusions or make decisions based on the information that is
available.
3. User interface: The user interface is the part of the expert system that allows
users to interact with the system and input data or ask questions

There are several different types of knowledge that are used in knowledge-based
systems, including:

1. Factual knowledge: Factual knowledge is information about specific facts or

events that are relevant to the domain of the expert system.
2. Rule-based knowledge is knowledge that is represented in the form of explicit
rules or statements about how the system should behave in certain situations.
3. Procedure-based knowledge: Procedure-based knowledge is knowledge that is
represented in the form of procedures or steps that describe how to perform a
specific task.
4. Model-based knowledge: Model-based knowledge is a representation of a system
or process that can be used to make predictions or draw conclusions about the
behavior of the system. It may include mathematical models, simulations, or
other types of representations.

3. Systém produkčných pravidiel

A production rule system, also known as a rule-based system is system that uses a set of
rules or "productions" to make decisions or draw conclusions based on the information
that is available
4. Dopredné a spätné reťazenie pravidiel

Forward chaining is an approach to reasoning that starts with the available data and uses
a set of rules to draw conclusions or make decisions based on that data.

Backward chaining is an approach to reasoning that starts with the goal and uses a set of
rules to work backwards to the data, gathering the necessary information along the way.

5. Inferenčná sieť a spôsoby jej prehľadávania

An inference network is a type of artificial intelligence (AI) system that represents
knowledge as a set of nodes and arcs in a directed graph. The nodes represent concepts
or variables, and the arcs represent the relationships between the concepts or variables.
An inference network is used to perform tasks that require reasoning and
decision-making, such as diagnosis, classification, and prediction.

There are several different ways to traverse an inference network, including depth-first
search, breadth-first search, and best-first search.

1. Depth-first search, Breadth-first search,Best-first search…

6. Spracovanie neurčitých znalostí

There are several different ways to handle uncertain knowledge in AI systems, including:

1. Probabilistic reasoning: It allows the system to represent the likelihood of different

events or outcomes and to make decisions or predictions based on that likelihood.
2. Fuzzy logic:It allows the system to represent the degree of membership of a
concept in a particular set, and to make decisions or predictions based on that
degree of membership.
3. Bayesian networks: They allow the system to represent the relationships between
different variables and to make predictions or decisions based on the probabilities
of those variables.
7. Základné prístupy k získavaniu znalostí pre BZ

1. Expert knowledge elicitation: Expert knowledge elicitation involves consulting with

experts in the domain to gather the necessary knowledge for the BN.
2. Data-driven learning: This may involve using machine learning algorithms to learn
the structure of the BN from data
3. Hybrid approaches: Hybrid approaches involve a combination of expert knowledge
elicitation and data-driven learning

8. Vysvetľovací mechanizmus
Explanatory mechanisms are important in AI because they allow users to understand
how the system arrived at a particular conclusion or decision.

9. Vyhodnocovanie ZS a jeho bezpečnosť

Evaluating a knowledge-based system (KBS) involves assessing its performance. This

may involve measuring the accuracy of the system's outputs, the speed of process. It may
also involve assessing the system's usability, reliability, and maintainability.

Evaluating a KBS is involves protecting the system and its data from unauthorized access
or modification.
10. Porovnanie ZS s ľudským riešiteľom a jeho aplikačné možnosti ZS

1. One major difference between a KBS and a human problem-solver is that a KBS
relies on a predetermined set of rules or knowledge, whereas a human
problem-solver can use a variety approaches.
2. KBS can process and analyze large amounts of data much faster. However, a KBS
may lack the ability to make intuitive or creative leaps.

Some potential applications for KBS include: medicine, engineering, and computer
systems,spam filtering, image recognition,financial planning, weather forecasting,
customer behavior prediction etc.
Strojové učenie
1. Definícia strojového učenia (SU)

Machine learning (ML) is a type of artificial intelligence (AI) that allows systems to learn
and improve their performance without being explicitly programmed.

2. Reprezentácia vstupov a výstupov

Numeric representation:This is often used for data that can be naturally represented as
numbers.

Boolean representation: Boolean representation involves encoding data as binary values

(true or false).

Categorical representation: Categorical representation involves encoding data as a set of

discrete categories.

Vector representation: Vector representation involves encoding data as a set of numerical

values arranged in a fixed-length vector.

3. Proces strojového učenia – jeho fázy

Defining the problem: The first step in the process of ML is to define the problem.
Gathering and preparing the data: The next step is to gather and prepare the data
Training the model:ML system is trained using a subset of the data (the training set).
Evaluating the model: it is evaluated using the remaining data (the test set).
Model tuning: If the model's performance is not up to the desired standards, you may
need to tune the machine learning model parameters to improve its performance.
Deploying the model: Once the model has been trained and evaluated, it can be deployed
in a production environment.
4. Delenie algoritmov SU

1. Supervised learning algorithms: Use labeled training data to learn a mapping from
input data to output labels.
2. Unsupervised learning algorithms: Use unlabeled data to learn patterns and
relationships in the data.
3. Reinforcement learning algorithms: Learn by interacting with an environment and
receiving feedback in the form of rewards or penalties
4. Deep learning algorithm:Use multiple layers of artificial neural networks to learn
complex relationships in the data
5. Transfer learning algorithms: Are a type of supervised learning algorithm that can
use knowledge learned from one task to improve performance on a related task.

5. Stromové modely - rozhodovacie stromy

Decision trees are a type of machine learning model that uses a tree-like structure to
make decisions based on input data.

To build a decision tree, an algorithm is used to split the data into smaller and smaller
subsets based on the values of the input features.Decision trees are commonly used for
classification tasks, although they can also be used for regression tasks.
6. Princípy na ktorých pracuje algoritmus ID3

The ID3 (Iterative Dichotomiser 3) algorithm is a decision tree learning algorithm that was
developed by Ross Quinlan in the 1980s. It is based on the concept of entropy, which measures
the uncertainty or randomness in a dataset. The ID3 algorithm works by iteratively dividing the
data into smaller and smaller subsets based on the values of the input features, with the goal of
creating leaf nodes that are as pure as possible.

7. Učenie súborom metód – Náhodné lesy

Ensemble methods involve training multiple models and combining their predictions to
improve the overall performance of the system.

8. SVM – Stroje podporných vektorov

Support Vector Machines (SVMs) are a type of supervised machine learning algorithm
that can be used for classification or regression tasks. The goal of an SVM is to find the
hyperplane in a space that maximally separates the classes in the training data.
9. Naivný Bayes klasifikátor

based on the Bayes theorem of probability, often used for classification tasks.

The Naive Bayes classifier works by making predictions based on the probability of an
event occurring given the presence of certain features. For example, if we want to classify
an email as spam or not spam, the Naive Bayes classifier would estimate the probability
of an email being spam given the presence of certain words or phrases in the email.
10. Chyba klasifikácie a miery efektívnosti

Classification error defined as the proportion of misclassified data points in a dataset. For
example, if a model correctly classifies 90 out of 100 data points, its classification error is
10%.

prof. Ing. Peter Sinčák, CSc.

Úvod o subsymbolickej UI
1. Čo je podstata subsymbolickej umelej inteligencie, forma znalostí – deklaratívne,
proceduralne , tacitne znalosti, znalosti v dátach , znalosti v ľudskej skúsenosti.

Subsymbolic AI often relies on statistical and probabilistic methods, and is typically used
for tasks that require learning from data or adapting to changing environments.

1. Declarative knowledge is knowledge that can be expressed in the form of facts or

statements.
2. Procedural knowledge: Procedural knowledge is knowledge of how to perform a
task or procedure.
3. Tacit knowledge: Tacit knowledge is knowledge that is difficult to express in words
or symbolic form. It is often based on experience.
4. Data-based knowledge: Data-based knowledge is knowledge that is derived from
data, such as patterns or trends in the data.
5. Human experience-based knowledge: Human experience-based knowledge is
knowledge that is based on the experiences of humans.

2. Priznakový priestor a jeho význam v S-UI, Aproximácia funkcie a vzťah k logike

In machine learning, a feature space is a set of all possible values that a feature can take
on.

Approximation of a function refers to the ability of a machine learning model to

approximate an unknown function that describes the relationship between the input data
and the output variable.

The relationship between function approximation and logic is that logic can be used to
encode rules or constraints that the function should satisfy.
3. Biologická motivácia subsymbolickej umelej inteligencie

Sub-symbolic AI seeks to replicate some of the possibilities using computational models

inspired by brain structure and function.

4. Black Box a White box prístupy v S-UI

In artificial intelligence and machine learning, a black box model is a model with internal workings that

are not visible to the user.

On the other hand, a "white box" model is a model with its inner workings visible to the user.

5. Explainable AI , Trusthworthy AI , Machine Trust Index

Explainable AI (XAI) focuses on developing and using algorithms and models that can
provide explanations for their predictions and decisions.

Trustworthy AI refers to the use of AI in a way that is ethical, responsible, and transparent,
and that does not harm or discriminate against individuals or groups.

The Machine Trust Index (MTI) is a measure of the level of trust that users have in a
particular AI system.

6. Brain like systemy, AI inferenčné počítače, AI učiace sa počítače

Brain-like systems, also known as artificial neural networks, are computational models
that are inspired by the structure and function of the human brain.

AI inference computers are computers that are specifically designed to perform tasks
related to artificial intelligence.

AI learning computers are computers that are able to learn from data and adapt to new
situations.
7. Základy tvorby človeka, zrozumiteľný systém na vstupe a výstupe, elementy neistoty v SUI ,
fuzzy množiny

The creation of a human-like artificial intelligence (AI) system would require natural
language processing, perception, reasoning, and decision-making.

AI system would need to have a understandable input and output system. This would
require the development of advanced natural language processing and speech
recognition capabilities.

There are several elements of uncertainty that can be present in AI systems, including:
Data uncertainty,Prediction uncertainty.

Fuzzy sets are characterized by a membership function, which defines the degree to
which an element belongs to the set.

8. Prírodne inšpirované algoritmy , prehľadanie, Kansei systémy, Chromozomy, fenotyp,

genotyp, prehľadávaci priestor, operácie „presunu” v priestore

Examples of nature-inspired algorithms include genetic algorithms and particle swarm

optimization.

Search algorithms are algorithms that are used to find a solution to a problem by
exploring a search space and evaluating potential solutions.

Kansei systems are systems that are designed to capture and analyze human emotions
and preferences in order to improve products or services.

Chromosomes are structures that are found in the cells of living organisms, and that
contain genetic information in the form of DNA.

The phenotype of an organism is the combination of its physical characteristics,

behaviors, and traits that result from the interaction between its genetic makeup
(genotype) and the environment.

The search space of an algorithm is the set of all possible solutions to a problem, or the
set of all possible states that can be reached during the execution of an algorithm.

Operations that involve "movement" is actions moving solution to a different part of the
search space to explore other solutions.
9. Tri základné výzvy v UI – Klasifikácie, Predikcia, Zreťazené rozhodovanie s odmenou
a trestom
SK: Klasifikácia - proces zaraďovania údajov do špecifických skupín alebo tried.
Predikcia - proces použitia modelu na predikciu výsledku na základe vstupných údajov.
Rozhodovanie s odmenou a trestom - proces trénovania modelov na prijímanie rozhodnutí,
ktoré maximalizujú odmenu a minimalizujú trest.
ENG: Classification - process of assigning data to specific groups or classes.
Prediction - process of using a model to predict an outcome based on input data.
Decision making with reward and punishment - process of training models to make
decisions that maximize a reward and minimize a punishment.

Neurónové siete
19. Biologická inšpirácia neurónových sietí
SK: Biologická inšpirácia neurónových sietí - koncept z neurofyziológie na vytvorenie
umelej neurónovej siete podobnej ľudskej sieti v mozgu.
ENG: The biological inspiration of neural networks - concept from neurophysiology to
create an artificial neural network similar to the human network in the brain.

20. Topológia neurónových sietí

SK: Topológia neurónových sietí - usporiadanie a komunikácia neurónov v rámci siete. V
umelých neurónových sieťach topológia siete určuje, ako sú neuróny navzájom prepojené a ako
v sieti prúdia informácie.
Rôzne topológie sa používajú pre rôzne typy úloh. Výber správnej topológie pre konkrétnu
úlohu je dôležitou súčasťou návrhu a tréningu neurónových sietí.
ENG: The topology of neural networks - arrangement and communication of neurons
within the network. In artificial neural networks, the network topology determines how neurons
are connected to each other and how information flows through the network.
Different topologies are used for different types of tasks. Choosing the right topology for a
particular task is an important part of neural network design and training.
21. Kontrolované verzus – nekontrolované učenie NS
SK: Kontrolované učenie - učenie, pri ktorom model dostane správne výstupné značky a
učí sa predpovedať tieto značky pre nové vstupné údaje.
Nekontrolované učenie - učenie, pri ktorom model musí odhaliť vzory v údajoch a zoskupiť
údaje do kategórií bez toho, aby poskytol správne výstupné štítky.
ENG: Supervised learning - learning, where the model is given the correct output labels
and learns to predict those labels for new input data.
Unsupervised learning - learning where the model must detect patterns in the data and group
the data into categories without giving the correct output labels.

22. Príznakový priestor a jeho formy ako vstup do NS

IThere are several different forms that a feature space can take, depending on the type of
data and the goals of the analysis.

Numeric feature space: This type of feature space represents data as a set of numerical
values.

Categorical feature space: This type of feature space represents data as a set of
categories or discrete values.

Binary feature space: This type of feature space represents data as a set of binary values,
such as 0 or 1, or true or false.

Multivariate feature space: This type of feature space represents data as a set of multiple
variables or features.

23. Výstup neurónovej siete a jeho formy

1. Classification: The output is a class label, indicating which category the input data
belongs to.
2. Regression: output of is a numerical value.
3. Probability: output a probability value
4. Activation value: The output of is often referred to as its activation value.
24. Perceptrón ako jednoduchá neurónová sieť

A perceptron is a type of simple neural network that was developed in the 1950s as a
model of how the brain works. It consists of a single layer of artificial
neurons.Perceptrons are capable of learning simple linear classifiers.

25. Metoda spätného sírenia chyby – základný princíp

Network is presented with input data and corresponding output labels, and the network's
parameters (weights and biases) are adjusted to minimize the error between the
predicted output and the true output.
26. Univerzálna aproximačná teoréma

The universal approximation theorem is a mathematical result that a neural network with
a single hidden layer can approximate any continuous function to any desired degree of
accuracy, given enough hidden units.

27. Deep Learning ako komplex neurónových podsietí

Deep neural networks are composed of multiple layers of interconnected artificial
neurons, and each layer processes and transforms the input data in a different way. The
first layer receives the raw input data, and the subsequent layers process the data by
learning and extracting increasingly abstract and complex features. The final layer of the
network produces the output, which can be used for prediction, classification, or other
tasks.
28. Aplikačný potenciál neurónových sietí – kedy ich použiť

Image recognition: Neural networks can be trained to recognize and classify images.

Speech recognition: Neural networks can be used to transcribe and translate spoken
language.

Machine translation: Neural networks can be used to translate text from one language to
another.

Predictive modeling: Neural networks can be used to predict future outcomes.

Robotics: Neural networks can be used to control robots and other devices.
Evolučné algoritmy
10. Čo je základný princíp evolučných algoritmov ?

The basic principle of evolutionary algorithms is to generate a population of candidate

solutions to a problem, evaluate the fitness of each solution, and select the fittest
solutions to reproduce and create a new generation of solutions.

11. Základná bloková schéma evolučného algoritmu

Initialization: The algorithm begins by generating a population of candidate solutions to

the problem, known as individuals.

Evaluation: The fitness of each individual is evaluated according to a fitness function,

which measures the quality of the solution.

Selection: The fittest individuals are selected from the population based on their fitness,
and they are chosen to reproduce and create a new generation of solutions.

Reproduction: Genetic operators, such as crossover and mutation, are applied to the
selected individuals to generate new solutions by combining and modifying their traits.

Replacement: The new generation of solutions replaces the previous generation, and the
process is repeated iteratively until the algorithm converges on an optimal solution, or
until a predetermined number of generations has been reached.
12. Aké sú základné pojmy v EA ?

1. Population is a set of candidate solutions to a problem.

2. Individual is a representation of a candidate solution to a problem.
3. Fitness function: The fitness function is a measure of the quality of a solution.
4. Selection is the process of choosing the fittest individuals from the population to
reproduce and create a new generation of solutions.
5. Reproduction is the process of generating new solutions by combining and
modifying the traits of the fittest individuals.
6. Crossover is a genetic operator that combines the traits of two parent solutions to
create a new offspring.
7. Mutation is a genetic operator that introduces random changes to the traits of an
individual.
8. Replacement is the process of replacing the previous generation of solutions with
the new generation.

13. Aké typy Genetických operátorov poznáme ?

Crossover, Mutation, Inversion , Scramble, Translation, Gene recombination, Gene

transposition

14. Vedeli by sme využiť GA v učení neurónových sietí ?

Yes, evolutionary algorithms (EAs) can be used to learn the weights and biases of neural
networks.
15. Aký je vzťah medzi GA a simuláciou kreativity ?
One way in which EAs can simulate creativity is by generating novel and diverse solutions
to a problem.
Another way in which EAs can simulate creativity is by combining and modifying existing
solutions to a problem in novel and interesting ways.
16. Čo je Genetické Programovanie ? Čo je výsledok GP ?

The goal of GP is to find a program that satisfies the given problem specification, or a set
of constraints and objectives that the program must satisfy. The result of GP is a program
that solves the given problem or performs the given task.
17. Čo sú Interaktívne Evolučné výpočty – uveďte príklad

Interactive evolutionary computation (IEC) is a type of evolutionary algorithm that involves

human interaction in the evolution process.

Example of IEC is IEM. IEM involves a human listener who assesses the quality of the
generated music and provides feedback to guide the evolution process.

18. Aplikačný potenciál Evolučných výpočtov

Optimization of manufacturing processes, such as production scheduling, inventory

management, and quality control

Design of engineering systems, such as aircraft, bridges, and automobiles

Analysis and prediction of financial markets, such as stock prices, exchange rates, and
risk management

Generation and optimization of art and music, such as paintings, sculptures, and
melodies

Evolution of game strategies, such as chess, poker, and go