| 1 | State-of-the-art flash memory devices and post-flash emerging memories显示文摘Although conventional Floating gate (FG) flash memory has recently gone into the 2X nm node, the technology challenges are formidable below 20 nm. Charge-trapping (CT) devices are promising to scale beyond 20 nm but below 10 nm both CT and FG devices hold too few electrons for robust MLC (multi-level cell, or more than one bit storage per cell) storage. However, due to the simpler structure and its more robust storage (not sensitive to tunnel oxide defects since charges are stored in deep trap levels), CT is much more desirable than FG in 3D stackable Flash memory. Optimistically, 3D CT Flash memory may allow the density increase to continue for at least another decade beyond the 1Xnm node. In this paper, we review the current status of FG devices, their scaling challenges, and the operation principles of CT devices and several variations such as MANOS and BE-SONOS. We will then discuss various 3D memory architectures, technology challenges and address the poly-silicon thin film transistor (TFT) issues. Devices that do not rely on charge storage are naturally not limited by the number of electrons, thus promise further scaling below 10 nm. Several of the most promising post-flash era devices, their operation principle and critical issues are reviewed. (One of them, phase change memory, will be covered in a separate article thus not included here.) Their potential applications and challenges for 3D stacking are critically examined. | LU ChihYuan LUE HangTing | 2011 | Science China(Information Sciences)2011，54，5： | 3 |
| 4 | Deep Scattering Spectra with Deep Neural Networks for Acoustic Scene Classification Tasks显示文摘As one of the most commonly used features,Mel-frequency cepstral coefficients(MFCCs)are less discriminative at high frequency.A novel technique,known as Deep scattering spectrum(DSS),addresses this issue and looks to preserve greater details.DSS feature has shown promise both on classification and recognition tasks.In this paper,we extend the use of DSS feature for acoustic scene classification task.Results on Detection and classification of acoustic scenes and events(DCASE)2016 and 2017 show that DSS provided 4.8%and 17.4%relative improvements in accuracy over MFCC features,within a state-of-the-art time delay neural network framework. | ZHANG Pengyuan CHEN Hangting BAI Haichuan YUAN Qingsheng | 2019 | Chinese Journal of Electronics2019，28，6： | 1 |
您所在的机构暂未开通此部分的功能，请联系管理员