FigureAsia 35 Under 35 · Science
Zhengwu Liu
Age 29 · Memristor brain–computer interfaces · China / Hong Kong
First author of a 2025 adaptive memristor decoder that co-evolves with changing brain signals.
- Approximate age at the edition eligibility date
- 29
- Field
- Neuroelectronics
- Country or region
- China / Hong Kong
- FigureAsia U35 Assessment
- 79.6 / 100
Profile
Career and documented record
Brain–computer interfaces drift because neural signals change while a fixed decoder does not. Zhengwu Liu first-authored a 2025 Nature Electronics study using a 128,000-cell memristor chip as an adaptive neuromorphic decoder that updates alongside the user.
In an extended interaction task involving ten participants, the paper reported roughly 20% higher decoding accuracy than an interface without co-evolution. A real-time brain-controlled drone demonstration showed that the hardware and learning loop could operate together rather than as separate simulations.
The participant study is small and not a medical trial. Its scientific contribution is the integration of adaptive memory hardware with non-stationary brain signals—a problem central to making interfaces reliable over time.
FigureAsia selection
Why Zhengwu Liu is on the list
Liu is selected for a 2025 result that joins device physics, learning and human interaction in one complete system. First authorship and a clear measured advantage make his agency and contribution unusually legible.
Verified work
The 2025–26 record
Adaptive memristor decoder
First-authored a 128k-cell neuromorphic decoder for brain–computer interfaces.
Ten-participant co-evolution test
Reported about 20% higher accuracy than a non-co-evolving comparison.
Real-time drone control
Demonstrated online decoding in a brain-controlled flight task.
Field context
The work in its field
BCI decoders face a moving target: neural activity changes with attention, learning and electrode conditions. Adaptation must improve accuracy without becoming unstable or opaque.
FigureAsia U35 Assessment
Assessment breakdown
79.6out of 100
Substantive 2025–2026 contribution
14.2 / 20
First-authored a 128k-cell neuromorphic decoder for brain–computer interfaces.
Verified scientific impact
11.5 / 15
Nature Electronics publication and a human interaction study provide a strong early system-level record.
Originality and distinction
8.3 / 10
The distinction lies in co-locating adaptive learning in memristor hardware so decoder and user can evolve together.
Field influence
8 / 10
The contribution gives memristor brain–computer interfaces a new method, limit or line of argument with relevance beyond one paper.
Individual agency
8.3 / 10
Liu is first author and an HKU lead contributor; senior and collaborating laboratories retain explicit credit.
Durability and trajectory
4.2 / 5
The record shows continuity at University of Hong Kong: this contribution belongs to a wider, sustained agenda.
Asian significance and global relevance
4.4 / 5
Chinese engineer educated at UESTC and Tsinghua and now leading neuroelectronic research in Hong Kong.
Evidential validity and reproducibility
6.4 / 8
The comparison is measured in ten participants and a real-time task; clinical generalisation is not claimed.
Advance in scientific knowledge
5.9 / 7
The work shows that neuromorphic plasticity can track non-stationary neural signals in an interactive loop.
Translational or methodological utility
4.1 / 5
Adaptive, energy-efficient decoders could support more practical non-invasive and future implantable interfaces.
Responsible research stewardship
4.3 / 5
Privacy, security and long-duration human safety remain explicit conditions on translation.