Neural Circuit Verification

Assisted by in-house AI deep compression toolchain (ezLabel, ezModel, ezQUANT, ezHybrid-M), the proposed technology supports automatic AI model design and optimization with the integrated performance of 120x model size reduction and 70x power reduction in 2D CNN model, and develops a world-first 1/2/4/8-bit CNN model realized by the developed high efficiency Hybrid fixed point CNN NPU (Hybrid-NPU), which has been verified in Xilinx ZCU102 FPGA and achieves the performance up to 2.5 TOPS(8-b)/ 20TOPS(1-b)＠28nm technology running at 550MHz and 4TOPS/W energy efficiency.

This project integrates clinical physical and mental medical practice, medical engineering, information engineering and social welfare units, adopts cross-fields and uses AI technology, and builds an innovative artificial intelligence auxiliary evaluation and treatment system for dementia care based on cultural context. Establish an early intervention mechanism for early detection of dementia, and establish measures to improve cognitive function, emotional state, social and daily behavioral functions.

The SPC process technology developed by our team has a patented cloth surface treatment and multiple layers of conductive circuit layers (Layout) without etching and transfer printing processes. It can not only be continuously automated production (piece to piece), but also Environmental protection, in line with the green trend of cleaner production.

The technology uses cathodic arc deposition combined with electromagnet field control and optical emission spectroscopy (OES) for the deposition of new functional high entropy alloy (HEA) nitride coatings. The HEA nitride coatings possessed such high hardness and excellent hydrophobicity with low surface energy which are beneficial for high speed dry cutting. Finally, the coatings were deposited on the micro end mills for cutting test of printed circuit board. The HEA ACZN coated tools can possess high wear resistance and outperform the uncoated tools at least 4 times of tool lives.

Machines need annotations to learn, but human babies learn human languages with almost no annotations. Can machines do the same thing? To allow machines to learn human languages with only observations like human babies, a research team at Taiwan has partnered with the speech research groups in Meta, CMU, MIT, and JHU to develop a brand new self-supervised speech processing evaluation framework, Speech Processing Universal PERformance Benchmark (SUPERB).