Jiri Kubalik

Dr. Jiri Kubalik

orcid

Research Focus & Metrics

Academic Career Length

8Years (2019-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
7 March 2024 in IEEE Access
Verified
Journal Article
Open access
MVMartin Vastl
Martin Vastl
JKJonáš Kulhánek
Jonáš Kulhánek
JKJiri Kubalik
Jiri Kubalik
EDErik Derner
Erik Derner
RBRobert Babuška
Robert Babuška
Many real-world systems can be naturally described by mathematical formulas. The task of automatically constructing formulas to fit observed data is called symbolic regression. Evolutionary methods such as genetic programming have been commonly used to solve symbolic regression tasks, but they have significant drawbacks, such as high computational complexity. Recently, neural networks have been applied to symbolic regression, among which the transformer-based methods seem to be most promising. After training a transformer on a large number of formulas (in the order of days), the actual inference, i.e., finding a formula for new, unseen data, is very fast (in the order of seconds). This is considerably faster than state-of-the-art evolutionary methods. The main drawback of transformers is that they generate formulas without numerical constants, which have to be optimized separately, yielding suboptimal results. We propose a transformer-based approach called SymFormer, which predicts the formula by outputting the symbols and the constants simultaneously. This helps to generate formulas fitting the data more accurately. In addition, the constants provided by SymFormer serve as a good starting point for subsequent tuning via gradient descent to further improve the model accuracy. We show on several benchmarks that SymFormer outperforms state-of-the-art methods while having faster inference.

8 October 2021 in IEEE Access
Verified
Journal Article
Open access
JKJiri Kubalik
Jiri Kubalik
EDErik Derner
Erik Derner
JZJan Zegklitz
Jan Zegklitz
RBRobert Babuska
Robert Babuska
Reinforcement learning algorithms can solve dynamic decision-making and optimal control problems. With continuous-valued state and input variables, reinforcement learning algorithms must rely on function approximators to represent the value function and policy mappings. Commonly used numerical approximators, such as neural networks or basis function expansions, have two main drawbacks: they are black-box models offering little insight into the mappings learned, and they require extensive trial and error tuning of their hyper-parameters. In this paper, we propose a new approach to constructing smooth value functions in the form of analytic expressions by using symbolic regression. We introduce three off-line methods for finding value functions based on a state-transition model: symbolic value iteration, symbolic policy iteration, and a direct solution of the Bellman equation. The methods are illustrated on four nonlinear control problems: velocity control under friction, one-link and two-link pendulum swing-up, and magnetic manipulation. The results show that the value functions yield well-performing policies and are compact, mathematically tractable, and easy to plug into other algorithms. This makes them potentially suitable for further analysis of the closed-loop system. A comparison with an alternative approach using neural networks shows that our method outperforms the neural network-based one.

19 June 2023 in IEEE Access
Verified
Journal Article
Open access
JKJiri Kubalik
Jiri Kubalik
EDErik Derner
Erik Derner
RBRobert Babuška
Robert Babuška
Many real-world systems can be described by mathematical models that are human-comprehensible, easy to analyze and can be helpful in explaining the system’s behavior. Symbolic regression is a method that can automatically generate such models from data. Historically, symbolic regression has been predominantly realized by genetic programming, a method that evolves populations of candidate solutions that are subsequently modified by genetic operators crossover and mutation. However, the approach suffers from several deficiencies: it does not scale well with the number of variables and samples in the training data – models tend to grow in size and complexity without an adequate accuracy gain, and it is hard to fine-tune the inner model coefficients using just genetic operators. Recently, neural networks have been applied to learn the whole analytic model, i.e., its structure as well as the coefficients, by means of gradient-based optimization algorithms. In this paper, we propose a novel neural network-based symbolic regression method that constructs physically plausible models based on very small training data sets and prior knowledge about the system. The method employs an adaptive weighting scheme to effectively deal with multiple loss function terms and an epoch-wise learning process to reduce the chance of getting stuck in poor local optima. Furthermore, we propose a parameter-free method for choosing the model with the best interpolation and extrapolation performance out of all models generated throughout the whole learning process. We experimentally evaluate the approach on four test systems: the TurtleBot 2 mobile robot, the magnetic manipulation system, the equivalent resistance of two resistors in parallel, and the longitudinal force of the anti-lock braking system. The results clearly show the potential of the method to find parsimonious models that comply with the prior knowledge provided.

21 March 2019 in Sensors
Verified
Journal Article
Open access
PDF file
MKMiroslav Kulich
Miroslav Kulich
JKJiri Kubalik
Jiri Kubalik
LPLibor Přeučil
Libor Přeučil
This paper deals with the problem of autonomous navigation of a mobile robot in an unknown 2D environment to fully explore the environment as efficiently as possible. We assume a terrestrial mobile robot equipped with a ranging sensor with a limited range and 360 ° field of view. The key part of the exploration process is formulated as the d-Watchman Route Problem which consists of two coupled tasks—candidate goals generation and finding an optimal path through a subset of goals—which are solved in each exploration step. The latter has been defined as a constrained variant of the Generalized Traveling Salesman Problem and solved using an evolutionary algorithm. An evolutionary algorithm that uses an indirect representation and the nearest neighbor based constructive procedure was proposed to solve this problem. Individuals evolved in this evolutionary algorithm do not directly code the solutions to the problem. Instead, they represent sequences of instructions to construct a feasible solution. The problems with efficiently generating feasible solutions typically arising when applying traditional evolutionary algorithms to constrained optimization problems are eliminated this way. The proposed exploration framework was evaluated in a simulated environment on three maps and the time needed to explore the whole environment was compared to state-of-the-art exploration methods. Experimental results show that our method outperforms the compared ones in environments with a low density of obstacles by up to 12.5 % , while it is slightly worse in office-like environments by 4.5 % at maximum. The framework has also been deployed on a real robot to demonstrate the applicability of the proposed solution with real hardware.

17 April 2023 in Applied Sciences
Verified
Journal Article
Open access
PDF file
JKJiri Kubalik
Jiri Kubalik
LKLukáš Kurilla
Lukáš Kurilla
PKPetr Kadera
Petr Kadera
The facility layout problem is one of the fundamental production system management problems. It has a significant impact on overall system efficiency. This paper introduces a new facility layout problem that allows for choosing from multiple variants of each facility. The need for choosing the most suitable selection from the facility variants while at the same time optimizing other layout quality indicators represents a new optimization challenge. We build on our previous work where single- and multi-objective evolutionary algorithms using indirect representation were proposed to solve the facility layout problem. Here, the evolutionary algorithms are adapted for the problem of facility variants, including the new solution representation and variation operators. Additionally, a cooling schedule, whose role is to control the exploration/exploitation ratio during the course of the optimization process, is proposed. It was inspired by the cooling schedule used in the simulated annealing technique. The extended evolutionary algorithms have been experimentally evaluated on two data sets, with and without the alternative variants of facilities. The obtained results demonstrate the capability of the extended evolutionary algorithms to solve the newly formulated facility layout problem efficiently. It also shows that the cooling schedule improves the convergence of the algorithms.

20 October 2022 in arXiv
Verified
Preprint
Open access
PDF file
MVMartin Vastl
Martin Vastl
JKJonáš Kulhánek
Jonáš Kulhánek
JKJiri Kubalik
Jiri Kubalik
EDErik Derner
Erik Derner
RBRobert Babuška
Robert Babuška
Many real-world problems can be naturally described by mathematical formulas. The task of finding formulas from a set of observed inputs and outputs is called symbolic regression. Recently, neural networks have been applied to symbolic regression, among which the transformer-based ones seem to be the most promising. After training the transformer on a large number of formulas (in the order of days), the actual inference, i.e., finding a formula for new, unseen data, is very fast (in the order of seconds). This is considerably faster than state-of-the-art evolutionary methods. The main drawback of transformers is that they generate formulas without numerical constants, which have to be optimized separately, so yielding suboptimal results. We propose a transformer-based approach called SymFormer, which predicts the formula by outputting the individual symbols and the corresponding constants simultaneously. This leads to better performance in terms of fitting the available data. In addition, the constants provided by SymFormer serve as a good starting point for subsequent tuning via gradient descent to further improve the performance. We show on a set of benchmarks that SymFormer outperforms two state-of-the-art methods while having faster inference.