Toward practical applications of near-term quantum computers
Recent advancements in experiments have significantly improved the quality of quantum devices, creating new opportunities and challenges. On one hand, these devices allow us to explore physics that was previously inaccessible due to our lack of access to large Hilbert spaces. On the other hand, achieving fault-tolerant quantum computation, where ideal quantum algorithms can be implemented, still requires significant advances in quantum error correction. In this talk, I will present our recent efforts to develop novel machine learning methods aimed at advancing both fronts. To enable practical applications of current quantum devices, we introduced the Quantum Attention Network (QuAN), a model inspired by large language models and tailored for quantum data. QuAN overcomes the challenges posed by noise in quantum devices, enabling us to investigate many-body phenomena. Using QuAN, we observed distinct entanglement scalings in a driven hardcore Bose-Hubbard model, detected quantum complexity growth in random circuit dynamics, and learned mixed-state topological order in a toric code under both coherent and incoherent noise, shedding light on decoding thresholds. I will also discuss our work on improving practical quantum error correction. Motivated by recent experimental progress in error-corrected logical algorithms, we developed a machine learning framework for decoding logical circuits beyond quantum memory. This method achieves competitive accuracy with significantly lower computational costs compared to conventional approaches.
Livestream the event on Zoom here (Yale login required)