Academia-Industry Matching Event (AIME24)

21 Nov 2024, 08:00 → 22 Nov 2024, 14:00 Europe/Budapest

Mercure Budapest Castle Hill Hotel

HUN-REN Wigner RCP, together with the HEPTECH Network, is organizing the next

Academia-Industry Matching Event (AIME24)

in the topics

Artificial Intelligence, HPC and Quantum Computing

The aim of this event is to bring together Academic researchers and Industry experts to share ideas, potential applications, and foster collaborations in the field of theoretical and practical aspects of Artificial Intelligence, High Performance Computing (HPC) and Quantum Computing.

Patron

• Roland Jakab (HUN-REN, CEO)
• Charaf Hassan (BME, Rector)

Confirmed keynote speakers

• Matthew Schwartz (Harvard Univ.)
• Tibor Höltz (Furukawa Electric)
• Robert Izsak (Riverlane LTD)
• Enrique Rico Ortega (CERN)
• Ernesto Bonomi (Groq Inc.)
• Arturo Garcia (Barcelona Supercomputing Center)

Contributions to the conference are welcome.

Abstracts must contain a title shorter than 100 characters, the name and affiliation of the presenter and coauthors, and a maximum of 4000 characters of body text. Images should be sent separately from the text as the abstract will be reprocessed for display on the website.

Talk submission deadline: November 15, 2024

The call for abstracts is open.

Thursday, 21 November

09:00 → 09:30 Opening: Welcome & Opening

Conveners: Andras Telcs, Peter Levai (WIGNER RCP)

09:00 Welocme and Opening

Speaker: Roland Jakab (HUN-REN)

09:30 → 11:00 Artificial Intelligence

Convener: Peter Levai (WIGNER RCP)

09:30 Quantum Communication Research at the Budapest University of Technology and Economics

Speaker: Dr Eszter Udvary (Budapest University of Technology and Economics)

1_02_BME_GUE_v2.pdf

09:50 MILAB - Synergies between academic research and industrial needs

Speaker: Gábor Érdi-Krausz (HUN-REN SZTAKI)

AI_National_Lab_Hungary_021Nov2024_AIME.pptx

10:25 A Novel Approach to Intelligent Systems

Mainstream artificial intelligence (AI) solutions, while achieving considerable success in areas such as classification, generation, and natural language understanding, still face several notorious, long-standing challenges. These include unexpected classification errors, hallucinations in generative models, and catastrophic forgetting, among others.

Addressing these issues requires rethinking our understanding of intelligence itself. Current AI solutions resemble intuitive, automatic responses more than true "thinking". True thinking involves structured knowledge representation, incorporating concepts such as relevant features, governing principles, abstraction, and generalization.

In this talk, I will explore our perspective on these challenges, outlining the logic and implementation of an algorithm designed as an initial step toward deeper, more thoughtful AI. This approach aims to address core limitations in current systems and lay a foundation for more robust intelligent systems.

Speaker: Antal Jakovác (WIGNER RCP)

1_04_AIME2024.pdf

11:00 → 11:30 Coffee Break

11:30 → 13:00 Artificial Intelligence

Convener: Gergely Barnafoldi (Wigner RCP RMI of the Hungarian Academy of Sciences)

11:30 On the Causality Paradox in Large Language Models: The Known Unknowns

Large language models (LLMs) face challenges with complex
probabilistic, causal, and counterfactual reasoning, yet they
demonstrate promising capabilities in the automated construction of both
qualitative and quantitative probabilistic and causal models. This talk
provides an overview of current benchmarks for causal reasoning in LLMs,
methods for extracting knowledge from LLMs, fundamental limitations in
LLM inference processes, and recent approaches that integrate LLMs with
causal inference technologies.

Speaker: Péter Antal (Budapest University of Technology and Economics)

12:00 AI Applications with BOSCH

Speaker: Hunor István Lukács (Eötvös University)

AIME_Bosch_ELTE.pptx

12:30 Unified theory of temporal causal discovery

Temporally evolving systems are typically modeled by dynamic equations. A
key challenge in accurate modeling is understanding the causal relationships between subsystems, as well as identifying the presence and influence of unobserved hidden drivers on the observed dynamics. This paper presents a unified method capable of identifying fundamental causal relationships between pairs of systems, whether deterministic or stochastic. Notably, the method also uncovers hidden common causes among the observed variables. By analyzing the degrees of freedom in the system, our approach provides a more comprehensive understanding of both direct and indirect causal influences. This unified framework is validated through theoretical models and simulations, demonstratingits robustness and potential for broader application.

Speaker: Andras Telcs

13:00 → 14:00 Lunch

14:00 → 15:30 Quantum Computing and Technology

Convener: Zoltan Zimboras

14:00 Unlocking High-Performance AI Inference with Dense Tensor Computing on Groq's LPU

Speaker: Ernesto Bonomi (Groq Inc.)

14:30 On the learning abilities of photonic continuous-variable Born machines

We investigate photonic continuous-variable Born machines (CVBMs), which utilize photonic quantum states as resources for continuous probability distributions. Implementing exact gradient descent in the CVBM training process is often infeasible, bringing forward the need to approximate the gradients using an estimator obtained from a smaller number of samples, obtaining a quantum stochastic gradient descent (SGD) method. In this work, the ability to train CVBMs is analyzed using stochastic gradients obtained using relatively few samples from the probability distribution corresponding to homodyne measurement. The main obstacle to this analysis is that classically simulating CVBMs and obtaining samples is a demanding task, while a large number of iterations are needed to achieve convergence. The present research is enabled by a novel strategy to simulate homodyne detections of generic multimode photonic states using a classical computer. With this approach, a more comprehensive study of CVBMs is made possible, and the training of multimode CVBMs is demonstrated with parametric quantum circuits considerably larger than in previous articles. More specifically, we use the proposed algorithm to demonstrate learning of multimode quantum distributions using CVBMs. Moreover, successful CVBM trainings were demonstrated with the use of stochastic gradients.

Speaker: Zoltán Kolarovszki

15:00 AQMLator--An Auto Quantum Machine Learning E-Platform

Speaker: Tomasz Rybotycki (Center of Excellence in Artificial Intelligence, AGH University, Cracow)

15:30 → 16:00 Coffee Break

16:00 → 17:30 Quantum Computing and Technology

Convener: Zoltan Zimboras

16:00 Building a Quantum Computer Simulator based on Groq chips

Speaker: Peter Rakyta (Department of Physics of Complex Systems, Eötvös Loránd University)

rakyta_talk.pdf

16:30 Hybrid Quantum-Classical Reinforcement Learning in Latent Observation Spaces

Speaker: Daniel Nagy (HUN-REN Wigner Reasearch Centre for Physics)

AIME2024NagyDaniel.pdf

16:55 Theoretical high-energy physics and AI

Speaker: Matthew Schwartz (Harvard University)

AIME24-Schwartz.pdf

17:30 → 17:40 SciComp Presentation

Convener: Andras Telcs

18:00 → 20:00 Conference Dinner

Friday, 22 November

09:00 → 11:00 High-performance Computing

Convener: Örs Legeza (Wigner FK)

09:00 TBA

Speaker: Tibor Holtz (Furukawa LTD)

09:45 Quantum Phase Transitions: A Qutrit Perspective and Novel Order Parameter Discovery

Quantum phase transitions and symmetry-protected topological
phases remain at the forefront of quantum physics research. This
presentation highlights recent advancements in these fields, focusing on
two key contributions:
- Realizing the Haldane Phase with Qutrits: We
discuss the successful creation of the spin-1 Haldane phase on a qudit
quantum processor, as detailed in arXiv:2408.04702. This achievement,
using trapped-ion qutrits, provides experimental evidence for topological
features like long-range string order and spin fractionalization in spin-1
chains.
- Novel Order Parameter Discovery: We explore a groundbreaking
approach for identifying quantum phase transitions in systems lacking
conventional order parameters, as presented in arXiv:2408.01400. The
reduced fidelity susceptibility (RFS) vector field is employed to
construct phase diagrams and discover new order parameters, as exemplified
in the Axial Next Nearest Neighbor Interaction (ANNNI) Model. These
studies collectively demonstrate the power of quantum processors in
simulating complex quantum systems and expanding our understanding of
quantum phases of matter.

Speaker: Enrique Rico Ortega (CERN)

10:30 Tensor Network methods via AI accelerators

Speaker: Andor Menczer (ELTE, HUN-REN Wigner RCP)

11:00 → 11:30 Coffee Break

11:30 → 13:00 High-performance Computing

Convener: Örs Legeza (Wigner FK)

11:30 Advanced Tensor Network methods in HPC systems

Speaker: Artur Garcia (Barcelona Supercomputing Center)

12:15 Quantum Computing in Practice

In this contribution, I will briefly discuss the differences and
similarities between industrial and academic research and provide an
example of a joint project on magnetic systems and quantum computing.
Furthermore, I will summarize the most relevant research papers published
in our group on quantum computing, including those on embedding, the cost
analysis of pharmaceutical and solid-state algorithms, electron
correlation, the adaptation of classical techniques, such as the projector
augmented wave method, to quantum computers, first-quantization-based
methods and some of our work on early fault tolerant algorithms. Finally,
some conclusions will be drawn on the trends and prospects of quantum
computing in these application areas.

Speaker: Robert Izsak (Riverlane LTD)

13:00 → 13:10 Closing Remarks