Program
For your information, here is the current list of accepted speakers (in no particular order) and talk titles:
- Robert Huang - What cannot be learned in the quantum universe
- Spiros Kechrimparis - How Quantum Agents Can Change Which Strategies Are More Complex
- Magdalini Zonnios - Memory-minimal quantum generation of stochastic processes: spectral invariants of quantum hidden Markov models
- Seok Hyung Lie - Multipartite Quantum State Over Time and Markovianity
- Graeme Berke - Fundamental limits on quantum prediction
- Chengran Yang - An Efficient Quantum Algorithm for Discovering Rare Events of Stochastic Processes
- Josep Lumbreras - Learning finitely correlated states
- Giorgos Eftaxias - Advantages of multicopy nonlocality distillation and its application to minimizing communication complexity
- Rudy Raymond - Combinatorial Optimization with Few Qubits
- Leonardo Banchi - Learning, Compression and Quantum Information
- Daniel Park - Learning with Repeated Classical-Quantum Interactions
Abstracts
Combinatorial Optimization with Few Qubits
Abstract: The potential of quantum computing for Combinatorial Optimization (CO) is significant, but current quantum computers lack the necessary physical resources, particularly the number of qubits, to tackle problem instances on the same scale as classical computers. Various methods have been developed to scale problem instances that can be addressed with a limited number of qubits, showing promise. Here, we introduce a method for CO using few qubits, involving the application of Quantum Random Access Code (QRAC) for encoding variables into few qubits, Classical Shadow (CS) for decoding the variables efficiently, and Coordinate Descent (CD) for optimizing quantum circuits to generate desirable quantum states for CO instances. The effectiveness of the proposed method is demonstrated through experiments on parameterized quantum circuits for solving instances of NP-hard problems such as the minimum bisection.
Quantum adaptive agents with efficient long-term memories
Abstract: Central to the success of adaptive systems is their ability to interpret signals from their environment and respond accordingly - they act as agents interacting with their surroundings. Such agents typically perform better when able to execute increasingly complex strategies. This comes with a cost: the more information the agent must recall from its past experiences, the more memory it will need. Here we investigate the power of agents capable of quantum information processing. We uncover the most general form a quantum agent need adopt to maximise memory compression advantages, and provide a systematic means of encoding their memory states. We show these encodings can exhibit extremely favourable scaling advantages relative to memory-minimal classical agents, particularly when information must be retained about events increasingly far into the past.
What cannot be learned in the quantum universe
Abstract: Recent advances have significantly enhanced our understanding of what can be efficiently learned in the quantum universe. However, certain fundamental aspects remain resistant to efficient learning using known algorithms. This talk explores several fundamental properties—including time, causal structure, topological order, noise—and demonstrates how they can be provably hard to learn. These results stem from our recent work on how to construct random unitaries (with Ma, arxiv:2410.10116) and generate them in extremely low depth (with Schuster and Haferkamp, arXiv:2407.07754). Examining these unlearnable aspects of our world sheds light on the fundamental limits of scientific inquiry in the quantum realm.
Learning, Compression and Quantum Information
Abstract: The relationship between learning and compression is a fundamental and increasingly studied area in machine learning and information theory, which provides insight into generalization, model selection, and the limits of learnability. Recent research suggests that effective learning algorithms inherently compress the data they process, extracting meaningful patterns while discarding irrelevant information. On the other hand, quantum information tools enable the compression of classical data into quantum memories using fewer resources. In this talk I'll explore the foundational connections between learning and compression, and then show how quantum agents can be used to define more efficient learning models, paving the way for potential quantum advantages in learning tasks.
Learning with Repeated Classical-Quantum Interactions
Abstract: The interaction between classical and quantum systems lies at the heart of many emerging quantum machine learning models. These repeated and controlled interactions give rise to temporal structures that can be harnessed for learning and decision-making, with classical and quantum components complementing one another. In this talk, I will explore how such repeated classical-quantum interactions contribute to improving optimization, expressivity, generalization, and robustness to noise in quantum machine learning. Drawing inspiration from both physical processes and algorithmic design, I will share recent insights that point toward a more dynamic view of quantum learning.
Sponsors
