no code implementations • 2 Mar 2021 • Eren Kurshan, Hai Li, Mingoo Seok, Yuan Xie
Over the last decade, artificial intelligence has found many applications areas in the society.
no code implementations • 11 Sep 2020 • Zhewei Jiang, Jiangyi Li, Pavan K. Chundi, Sung Justin Kim, Minhao Yang, Joonseong Kang, Seungchul Jung, Sang Joon Kim, Mingoo Seok
We design the algorithms for those operations to achieve minimal computation complexity while matching or advancing the accuracy of state-of-art Brain-Computer-Interface sorting and movement decoding.
no code implementations • 22 Jun 2020 • Dewei Wang, Pavan Kumar Chundi, Sung Justin Kim, Minhao Yang, Joao Pedro Cerqueira, Joonsung Kang, Seungchul Jung, Sangjoon Kim, Mingoo Seok
Always-on artificial intelligent (AI) functions such as keyword spotting (KWS) and visual wake-up tend to dominate total power consumption in ultra-low power devices.
no code implementations • 22 Jul 2019 • Peiye Liu, Bo Wu, Huadong Ma, Mingoo Seok
Recent studies on automatic neural architectures search have demonstrated significant performance, competitive to or even better than hand-crafted neural architectures.
no code implementations • 18 Oct 2018 • Peiye Liu, Wu Liu, Huadong Ma, Tao Mei, Mingoo Seok
To transfer the knowledge of intermediate representations, we set high-level teacher feature maps as a target, toward which the student feature maps are trained.
no code implementations • ICLR 2018 • Tianchan Guan, Xiaoyang Zeng, Mingoo Seok
With the same amount of data storage, our model can train a bigger network having more weights, achieving 1% less test error than the conventional binary neural network learning model.
no code implementations • 15 Sep 2017 • Saarthak Sarup, Mingoo Seok
Understanding the memory capacity of neural networks remains a challenging problem in implementing artificial intelligence systems.
no code implementations • 15 Sep 2017 • Tianchan Guan, Xiaoyang Zeng, Mingoo Seok
This enables a device with a given storage constraint to train and instantiate a neural network classifier with a larger number of weights on a chip and with a less number of off-chip storage accesses.
no code implementations • 1 Jul 2015 • Daniel Martí, Mattia Rigotti, Mingoo Seok, Stefano Fusi
We also show that the energy consumption of the IBM chip is typically 2 or more orders of magnitude lower than that of conventional digital machines when implementing classifiers with comparable performance.