A5004227278- Scheduling and Optimization Algorithms
- Resource-Constrained Project Scheduling
- Explainable Artificial Intelligence (XAI)
- Non-Destructive Testing Techniques
- Ethics and Social Impacts of AI
- Regional Development and Management Studies
- Software Engineering Research
- Adversarial Robustness in Machine Learning
- Manufacturing Process and Optimization
- Oil and Gas Production Techniques
- Parallel Computing and Optimization Techniques
- Machine Learning in Healthcare
- Constraint Satisfaction and Optimization
- Construction Project Management and Performance
- Magnetic Properties and Applications
- Advanced Data Storage Technologies
- Logic, Reasoning, and Knowledge
- Water Systems and Optimization
- Software Testing and Debugging Techniques
- Advanced Manufacturing and Logistics Optimization
- BIM and Construction Integration
- Electric Motor Design and Analysis
- Optimization and Packing Problems
- Smart Grid Energy Management
- Power System Optimization and Stability
University of Zagreb
2014-2024
Faculty of Electrical Engineering and Computing in Zagreb
2017
In the last decade, with availability of large datasets and more computing power, machine learning systems have achieved (super)human performance in a wide variety tasks. Examples this rapid development can be seen image recognition, speech analysis, strategic game planning many more. The problem state-of-the-art models is lack transparency interpretability. thereof major drawback applications, e.g. healthcare finance, where rationale for model's decision requirement trust. light these...
Understanding black box models has become paramount as systems based on opaque Artificial Intelligence (AI) continue to flourish in diverse real-world applications. In response, Explainable AI (XAI) emerged a field of research with practical and ethical benefits across various domains. This paper highlights the advancements XAI its application scenarios addresses ongoing challenges within XAI, emphasizing need for broader perspectives collaborative efforts. We bring together experts from...
Artificial intelligence has become mainstream and its applications will only proliferate. Specific measures must be done to integrate such systems into society for the general benefit. One of tools improving that is explainability which boosts trust understanding decisions between humans machines. This research offers an update on current state explainable AI (XAI). Recent XAI surveys in supervised learning show convergence main conceptual ideas. We list real world with concrete impact. The...
With the increasing complexity of power system structures and penetration renewable energy, driven primarily by need for decarbonization, operation control become challenging. Changes are resulting in an enormous increase complexity, wherein number active points grid is too high to be managed manually provide opportunity application artificial intelligence technology system. For flow control, many studies have focused on using generation redispatching, load shedding, or demand side...
Last decade has seen major improvements in the performance of artificial intelligence which driven wide-spread applications. Unforeseen effects such massad-option put notion AI safety into public eye. is a relatively new field research focused on techniques for building beneficial humans. While there exist survey papers safety, lack quantitative look at being conducted. The aspect gives data-driven insight about emerging trends, knowledge gaps and potential areas future research. In this...
Artificial intelligence (AI) has been embedded into many aspects of people's daily lives and it become normal for people to have AI make decisions them. Reinforcement learning (RL) models increase the space solvable problems with respect other machine paradigms. Some most interesting applications are in situations non-differentiable expected reward function, operating unknown or underdefined environment, as well algorithmic discovery that surpasses performance any teacher, whereby agent...
Integrating large language model (LLM) agents within game theory demonstrates their ability to replicate human-like behaviors through strategic decision making. In this paper, we introduce an augmented LLM agent, called the private which engages in deliberation and employs deception repeated games. Utilizing partially observable stochastic (POSG) framework incorporating in-context learning (ICL) chain-of-thought (CoT) prompting, investigated agent’s proficiency both competitive cooperative...
The data mesh is a novel management concept that emphasises the importance of domain before technology.The still in early stages development and many efforts to implement use it are expected have negative consequences for organizations due lack technological guidelines best practices.To mitigate risk outcomes this paper proposes mask-mediatorwrapper architecture as driver.The provides set prefabricated configurable components provide basic functionalities which requires.This shows how two...
Scheduling is a family of combinatorial problems where we need to find optimal time arrangements for activities. in applications are usually notoriously hard solve exactly. Existing exact solving procedures, based on mathematical programming and constraint programming, make manually-tuned heuristic choices. These heuristics can be improved by machine learning. In this paper, apply the graph convolutional neural network from literature speeding up general branch&bound solver learning its...
Proactive-reactive scheduling is important in the situations where project collaborators need to coordinate their efforts. The coordination mostly achieved through combination of shared baseline schedule and deviation penalties. In this paper, we present an extension predictive Gantt chart proactive-reactive needs. It can be used track evolution relationship between dynamic static elements time. are evolving probability distributions due uncertainty revealed information. time-agreements...
This paper deals with the mediator–wrapper architecture and observes it in more modern aspects by relating to architectural quanta. It is an important pattern that enables a flexible modular opposition monolithic architectures for data source integration systems. identifies certain realistic concrete scenarios where underperforms. These issues are addressed extension of via mask component type. The detailed so can be reasoned about without prescribing programming language or paradigm but...
In this paper, the integration of a power system simulator and reinforcement learning (RL) tools frameworks is presented. A proposed framework easily applicable can serve as for further developing, training, benchmarking RL algorithms on more complex tasks control. The usage standard enables broad range state-of-the-art to be implemented with high performance, scalability, substantial code reuse. Also, design suitable scaling onto high-performance computing (HPC) clusters which significantly...
The future of computation is massively parallel and heterogeneous with specialized accelerator devices instruction sets in both edge- cluster-computing. However, software development bound to become the bottleneck. To extract potential hardware wonders, would have solve following problems: device mapping, capability discovery, parallelization, adaptation new ISAs, many others. This systematic complexity will be impossible manually tame for human developers. These problems need offloaded...
We shall have a hard look at ethics and try to extract insights in the form of abstract properties that might become tools. want connect games, talk about performance ethics, introduce curiosity into interplay between competing coordinating well-performing offer view possible developments could unify increasing aggregates entities. All this is under long shadow cast by computational complexity quite negative games. This analysis first step toward finding modeling aspects be used AI for...
Stochastic Resource Constrained Project Scheduling (SRCPS) is among the hardest combinatorial problems. Exact calculations of interesting measures, such as expected project duration and probability satisfying deadline, using known probabilities are in #P even for relaxed instances problem where resource constraints ignored. The most common approach to use substantial simulation evaluate candidate solutions. All work so far uses ad-hoc developed environments with prevalent a priori generated...
An impossibility theorem demonstrates that a particular problem or set of problems cannot be solved as described in the claim. Such theorems put limits on what is possible to do concerning artificial intelligence, especially super-intelligent one. As such, these results serve guidelines, reminders, and warnings AI safety, policy, governance researchers. These might enable solutions some long-standing questions form formalizing theories framework constraint satisfaction without committing one...