Yang Tian
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Yang Tian has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
Haichao Tang
School of Materials Science and Engineering State Key Laboratory of Silicon Materials Zhejiang University Hangzhou 310027 China
3 shared publications- Jianguo Lu
Zhejiang University
3 shared publications Jian Lu
Zhejiang Laboratory, Hangzhou, China
2 shared publicationsJiangming Fu
2 shared publicationsKequan Xia
2 shared publications- Qi Liu
Fudan University
2 shared publications
Latest Publications
- Open Access
- 12 December 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1)
Error-aware probabilistic training for memristive neural networks
Analog computing-in-memory devices leverage fundamental physical laws for computation, greatly enhancing energy efficiency. However, the stochastic characteristics of analog devices conflict with the deterministic weight update of the backpropagation algorithm (BP), limiting training performance. To overcome the algorithm-device mismatch, we propose an error-aware probabilistic update method (EaPU) that updates the weights based on a specified probability derived from device writing noise. Compared to BP, EaPU reduces the number of weight updates to <1‰ with minimal performance loss. Furthermore, we validate EaPU experimentally on a 180 nm memristor system for image denoising and super-resolution and simulate its performance on ResNet and Vision Transformers. Results confirm that EaPU training yields over 60% accuracy improvement, with ~50.54× and 13.23× lower training energy (and ~35.51× and 11.26× lower inference energy) compared to BP-based memristor training and MADEM, respectively. Moreover, EaPU-based memristor hardware reduces training energy by nearly 6 orders of magnitude compared to graphics processing units. Here, we present a promising approach to precisely and efficiently train analog device-based deep neural networks.
- 12 December 2025
- Published by Springer Nature in Nature Communications
- Vol. 16 (1)
Open AccessError-aware probabilistic training for memristive neural networks
Analog computing-in-memory devices leverage fundamental physical laws for computation, greatly enhancing energy efficiency. However, the stochastic characteristics of analog devices conflict with the deterministic weight update of the backpropagation algorithm (BP), limiting training performance. To overcome the algorithm-device mismatch, we propose an error-aware probabilistic update method (EaPU) that updates the weights based on a specified probability derived from device writing noise. Compared to BP, EaPU reduces the number of weight updates to <1‰ with minimal performance loss. Furthermore, we validate EaPU experimentally on a 180 nm memristor system for image denoising and super-resolution and simulate its performance on ResNet and Vision Transformers. Results confirm that EaPU training yields over 60% accuracy improvement, with ~50.54× and 13.23× lower training energy (and ~35.51× and 11.26× lower inference energy) compared to BP-based memristor training and MADEM, respectively. Moreover, EaPU-based memristor hardware reduces training energy by nearly 6 orders of magnitude compared to graphics processing units. Here, we present a promising approach to precisely and efficiently train analog device-based deep neural networks.
- Open Access
- 9 May 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (19) , eadv2312
Fully memristive spiking neural network for energy-efficient graph learning
Parallel and energy-efficient searching of the shortest paths on a large graph is challenging. Conventional methods commonly used are sequential and computing intensive, rendering them inadequate for addressing large-scale and real-time situations. Here, we propose a highly parallel, computation- and energy-efficient approach to shortest path–based graph learning based on an emerging memristor spiking neural network via algorithm-device codesign. The shortest path is obtained parallelly in nature using simultaneous spike traveling instead of arithmetic calculation, achieving extremely low time and space complexity. A nonlinear weight mapping approach is proposed to counterbalance the neuron intrinsic nonlinearity to guarantee accuracy to support large-scale graphs. The memristor hardware capability is experimentally demonstrated in unsupervised and supervised classification tasks. The estimated energy efficiency of 517.82 giga-traversal edges per second per watt outperforms field programmable gate arrays by three to four orders of magnitude, providing a pathway toward highly energy-efficient graph computing hardware.
- 9 May 2025
- Published by American Association for the Advancement of Science (AAAS) in Science Advances
- Vol. 11 (19) , eadv2312
Open AccessFully memristive spiking neural network for energy-efficient graph learning
Parallel and energy-efficient searching of the shortest paths on a large graph is challenging. Conventional methods commonly used are sequential and computing intensive, rendering them inadequate for addressing large-scale and real-time situations. Here, we propose a highly parallel, computation- and energy-efficient approach to shortest path–based graph learning based on an emerging memristor spiking neural network via algorithm-device codesign. The shortest path is obtained parallelly in nature using simultaneous spike traveling instead of arithmetic calculation, achieving extremely low time and space complexity. A nonlinear weight mapping approach is proposed to counterbalance the neuron intrinsic nonlinearity to guarantee accuracy to support large-scale graphs. The memristor hardware capability is experimentally demonstrated in unsupervised and supervised classification tasks. The estimated energy efficiency of 517.82 giga-traversal edges per second per watt outperforms field programmable gate arrays by three to four orders of magnitude, providing a pathway toward highly energy-efficient graph computing hardware.
- 7 February 2020
- Published by Wiley in Advanced Functional Materials
- Vol. 30 (12)
Highly Stable Lithium−Sulfur Batteries Promised by Siloxene: An Effective Cathode Material to Regulate the Adsorption and Conversion of Polysulfides
Designing an appropriate cathode is still a challenge for lithium–sulfur batteries (LSBs) to overcome the polysulfides shuttling and sluggish redox reactions. Herein, 2D siloxene nanosheets are developed by a rational wet‐chemistry exfoliation approach, from which S@siloxene@graphene (Si/G) hybrids are constructed as cathodes in Li‐S cells. The siloxene possesses corrugated 2D Si backbone with abundant O grafted in Si6 rings and hydroxyl‐functionalized surface, which can effectively intercept polysulfides via synergistic effects of chemical trapping capability and kinetically enhanced polysulfides conversion. Theoretical analysis further reveals that siloxene can significantly elevate the adsorption energies and lower energy barrier for Li+ diffusion. The LSBs assembled with 2D Si/G hybrid cathodes exhibit greatly enhanced rate performance (919, 759, and 646 mAh g−1 at 4 C with sulfur loading of 1, 2.9, and 4.2 mg cm−2, respectively) and superb durability (demonstrated by 1000 cycles with an initial capacity of 951 mAh g−1 and negligible 0.032% decay rate at 1 C with sulfur loading of 4.2 mg cm−2). It is expected that the study presented here may open up a new vision toward developing high‐performance LSBs with siloxene for practical applications.
- 7 February 2020
- Published by Wiley in Advanced Functional Materials
- Vol. 30 (12)
Highly Stable Lithium−Sulfur Batteries Promised by Siloxene: An Effective Cathode Material to Regulate the Adsorption and Conversion of Polysulfides
Designing an appropriate cathode is still a challenge for lithium–sulfur batteries (LSBs) to overcome the polysulfides shuttling and sluggish redox reactions. Herein, 2D siloxene nanosheets are developed by a rational wet‐chemistry exfoliation approach, from which S@siloxene@graphene (Si/G) hybrids are constructed as cathodes in Li‐S cells. The siloxene possesses corrugated 2D Si backbone with abundant O grafted in Si6 rings and hydroxyl‐functionalized surface, which can effectively intercept polysulfides via synergistic effects of chemical trapping capability and kinetically enhanced polysulfides conversion. Theoretical analysis further reveals that siloxene can significantly elevate the adsorption energies and lower energy barrier for Li+ diffusion. The LSBs assembled with 2D Si/G hybrid cathodes exhibit greatly enhanced rate performance (919, 759, and 646 mAh g−1 at 4 C with sulfur loading of 1, 2.9, and 4.2 mg cm−2, respectively) and superb durability (demonstrated by 1000 cycles with an initial capacity of 951 mAh g−1 and negligible 0.032% decay rate at 1 C with sulfur loading of 4.2 mg cm−2). It is expected that the study presented here may open up a new vision toward developing high‐performance LSBs with siloxene for practical applications.
- 1 January 2020
- Published by Elsevier in Nano Energy
- Vol. 67
A high strength triboelectric nanogenerator based on rigid-flexible coupling design for energy storage system
- 1 January 2020
- Published by Elsevier in Nano Energy
- Vol. 67
A high strength triboelectric nanogenerator based on rigid-flexible coupling design for energy storage system
- 22 August 2019
- Published by Royal Society of Chemistry (RSC) in Chemical Communications
- Vol. 55 (73) , 10960-10963
FeSe 2 /carbon nanotube hybrid lithium-ion battery for harvesting energy from triboelectric nanogenerators
FeSe 2 –carbon nanotubes hybrid lithium ion batteries, exhibiting excellent electrochemical performances, could withstand the high-voltage pulse to directly harvest electricity from triboelectric nanogenerators with high stability.
- 22 August 2019
- Published by Royal Society of Chemistry (RSC) in Chemical Communications
- Vol. 55 (73) , 10960-10963
FeSe 2 /carbon nanotube hybrid lithium-ion battery for harvesting energy from triboelectric nanogenerators
FeSe 2 –carbon nanotubes hybrid lithium ion batteries, exhibiting excellent electrochemical performances, could withstand the high-voltage pulse to directly harvest electricity from triboelectric nanogenerators with high stability.
- 27 August 2018
- Published by Springer Nature in Journal of Materials Science
- Vol. 54 (1) , 571-581
Highly conductive NiSe2 nanostructures for all-solid-state battery–supercapacitor hybrid devices
- 27 August 2018
- Published by Springer Nature in Journal of Materials Science
- Vol. 54 (1) , 571-581
Highly conductive NiSe2 nanostructures for all-solid-state battery–supercapacitor hybrid devices