# Ponedeljkov seminar računalništva in informatike - Arhiv

2024 | 2023 | 2022 | 2021 | 2020 | 2019 | 2018 | 2017 | 2016 | 2015 | 2014 | 2013 | 2012 |

1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 |

## petek, 17. maj 2024 Zorica STANIMIROVIĆ: A GVNS-based solution approach to the Uncapacitated Single Allocation p-hub Maximal Covering Problem

V ponedeljek, 20. maja 2024, bo ob **16.00** uri izvedeno

predavanje v okviru PONEDELJKOVEGA SEMINARJA RAČUNALNIŠTVA IN INFORMATIKE

Oddelkov za Informacijske znanosti in tehnologije UP FAMNIT in UP IAM.

ČAS/PROSTOR: **20. maj 2024** ob **16.00** v **FAMNIT-VP2**.

-----------------------------------------------------

PREDAVATELJICA: Zorica STANIMIROVIĆ

-----------------------------------------------------

**Zorica Stanimirović **PhD, is a Full Professor at the Department for Numerical Mathematics and Optimization, Faculty of Mathematics, University of Belgrade and Coordinator for International Cooperation of the same institution. She graduated from the Faculty of Mathematics in 2000, received a master's degree in 2004, and PhD in 2007 at the same institution. She has been the local coordinator for several international projects and a member of program or organizational boards of several national and international conferences. From 2010 to 2015, she was Vice-Dean for Science and Research at the Faculty of Mathematics. Her research areas include Mathematical Modeling, Combinatorial Optimization, Metaheuristics, and Hybrid Optimization Methods. Up to now, she has published over 120 publications that have been cited 1182 times, her h-index is 19 and her i10 index is 32. More information and a list of publications are available at http://www.matf.bg.ac.rs/p/zoricast/pocetna/.

-------------------------------------------------------------------------------------------------------------------------------------------------------------

NASLOV: **A GVNS-based solution approach to the Uncapacitated Single Allocation p-hub Maximal Covering Problem**

-------------------------------------------------------------------------------------------------------------------------------------------------------------

POVZETEK:

Hub covering problems represent extensions of classical covering location problems and they are widely studied in the literature, due to their theoretical and practical importance in location science. This study considers the Uncapacitated Single Allocation p-hub Maximal Covering Problem (USApHMCP) with binary coverage criterion. The USApHMCP considers a complete symmetric graph G=(N,E), where N represents a set of nodes, while E denotes a set of edges. Transportation costs per unit of flow and the flow demand for each origin-destination (O-D) pair i-j are given (i,j ϵ N). The goal of USApHMCP is to choose locations of p hubs from the set H ⊆ N, and to allocate non-hub nodes to hubs, such that the total covered flow between O-D pairs is maximized. The flow between an O-D pair is considered "covered" if the transportation costs are within the given maximum service distance (coverage radius). Each non-hub node is assigned to exactly one hub and the incoming and outgoing flows are sent only through that hub (single allocation scheme). A mixed integer formulation of the USApHMCP is presented and used within the framework of CPLEX solver on hub instances from the literature. As USApHMCP belongs to the class of NP-hard optimization problems, a solution approach based on General Variable Neighborhood Search (GVNS) heuristic is developed to tackle instances of larger problem dimensions. Two variants of GVNS heuristics are proposed, which use different procedures in the solution improvement phase: Sequential Variable Neighborhood Descent and Nested Variable Neighborhood Descent. The impact of these two procedures on overall GVNS performance is investigated through extensive computational experiments on standard hub instances from the literature. The obtained results indicate the efficiency of both GVNS variants, however, the one with Nested Variable Neighborhood Descent procedure was more successful with respect to both solution quality and running times. The results of GVNS on large-scale problem instances are also presented, showing the potential of GVNS when solving USApHMCP on realistic-size hub networks.

Seminar bo potekal v **angleškem ****jeziku** v **FAMNIT-VP2** s pričetkom **ob 16:00 uri**.

Vabljeni.

## ponedeljek, 6. maj 2024 Domen VAKE: URŠKA - Univerzitetne rešitve: Študentski Komunikacijski Agent

V ponedeljek, 6. maja 2024, bo ob **16.00** uri izvedeno

predavanje v okviru PONEDELJKOVEGA SEMINARJA RAČUNALNIŠTVA IN INFORMATIKE

Oddelkov za Informacijske znanosti in tehnologije UP FAMNIT in UP IAM.

ČAS/PROSTOR: **6. maj 2024** ob **16.00** v **FAMNIT-VP2**.

---------------------------------------

PREDAVATELJ: Domen VAKE

---------------------------------------

**Domen Vake** is a third-year PhD student at UP FAMNIT under the mentorship of Assoc. Prof. Branko Kavšek and Assoc. Prof. Jernej Vičič. His research interests and projects mostly concern machine learning, more specifically natural language processing with large language models.

----------------------------------------------------------------------------------------------------

NASLOV: **URŠKA - Univerzitetne rešitve: Študentski Komunikacijski Agent**

----------------------------------------------------------------------------------------------------

POVZETEK:

This presentation will provide an overview of large language models (LLMs), focusing on their architecture, capabilities, and the underlying technologies that power them. We will then explore a practical application of these models through a case study on URŠKA, a Retrieval-Augmented Generation (RAG) based communication agent. URŠKA is designed to assist university students by answering frequently asked questions about deadlines, academic policies, and more. The talk will highlight the implementation challenges, the integration of RAG with URŠKA, and its plans.

Seminar bo potekal v **angleškem ****jeziku** v **FAMNIT-VP2** s pričetkom **ob 16:00 uri**.

Vabljeni.

## sreda, 17. januar 2024 Domen ŠOBERL: Qualitative Reasoning in Artificial Intelligence — bridging the gap between machine learning and human reasoning

V ponedeljek, 22. januarja 2024, bo ob **16.00** uri izvedeno

predavanje v okviru PONEDELJKOVEGA SEMINARJA RAČUNALNIŠTVA IN INFORMATIKE

Oddelkov za Informacijske znanosti in tehnologije UP FAMNIT in UP IAM.

ČAS/PROSTOR: **22. januar 2024** ob **16.00** v **FAMNIT-VP3**.

------------------------------------------

PREDAVATELJ: Domen ŠOBERL

------------------------------------------

**Domen Šoberl** received his PhD in computer science in 2021 from University of Ljubljana, Faculty of Computer and Information Science. He is currently employed as a teaching assistant at UP FAMNIT. His current research interests lie in various areas of artificial intelligence, including deep learning, generative adversarial networks, reinforcement learning, and qualitative reasoning.

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

NASLOV: **Qualitative Reasoning in Artificial Intelligence — bridging the gap between machine learning and human reasoning**

-------------------------------------------------------------------------------------------------------------------------------------------------------------------------

POVZETEK:

A fundamental difference between the way conventional methods of Artificial Intelligence (AI) make decisions and the way humans think and reason is that humans reason qualitatively, while AI typically makes decisions based on numerical computations. The gap between the two worlds — the human and the machine — becomes apparent when it comes to exchanging learned knowledge. Traditional numerical models usually consist of a large number of numerical parameters that convey little or no information to a human on why a particular decision or action was taken. On the other hand, it is very difficult for a human to describe their intuitive knowledge of how a particular mechanism works to an AI algorithm in a way that the algorithm can utilize in planning and decision-making. Qualitative Reasoning (QR) is a branch within AI research that focuses on how AI can reason about processes qualitatively, and present the findings in a form that approximates human intuition. I will present the historical origins of Qualitative Reasoning (QR) in AI, its later developments, and the current state of the art. I will focus on the area of agent learning and planning in continuous domains with numerical sensory and actuation systems. We will explore the full cycle of automated abstraction of qualitative representations from numerical observations, the search for symbolic solutions through qualitative reasoning, and the implementation of the found solutions in the original numerical domain. I will explain the foundations of qualitative physics and qualitative simulation, which is the basis for qualitative planning, and thus for predicting possible future behaviors in a symbolic and explainable way. I will present the results of experiments with different robot problems, such as learning to walk, learning to push objects, and learning to swing up and balance a pole.

Seminar bo potekal v **angleškem ****jeziku** v **FAMNIT-VP3** s pričetkom **ob 16:00 uri**.