The implementation of highly anticipated hardware neural networks (HNNs) hinges largely on the successful development of a low-power, high-density, and reliable analog electronic synaptic array. In this study, we demonstrate a two-layer Ta/TaO x /TiO2/Ti cross-point synaptic array that emulates the high-density three-dimensional network architecture of human brains. Excellent uniformity and reproducibility among intralayer and interlayer cells were realized. Moreover, at least 50 analog synaptic weight states could be precisely controlled with minimal drifting during a cycling endurance test of 5000 training pulses at an operating voltage of 3 V. We also propose a new state-independent bipolar-pulse-training scheme to improve the linearity of weight updates. The improved linearity considerably enhances the fault tolerance of HNNs, thus improving the training accuracy.

No abstract

No abstract

No abstract

The structure and formation mechanism of a microporous phase‐inversion poly(vinylidene fluoride) (PVDF) membrane exhibiting a relatively loosely packed agglomerate of semicrystalline globules are explored. The membrane has been prepared by the coagulation of a solution of PVDF in dimethylformamide by the action of 1‐octanol, which is a soft nonsolvent. Experimental observations pertain to the globule surface, which is dominated by a grainy nanostructure; the globular interior, which exhibits a range of fine structures (e.g., twisted sheets and treelike branches); and the globule–globule connections, which exhibit a sheetlike or ropelike structure. On the basis of the observed structural details and phase diagram considerations, it is proposed that the membrane structure is the result of a unique combination of a polymer crystallization and a liquid–liquid phase‐separation process, with end‐result globular structural features of remarkable uniformity. © 2003 Wiley Periodicals, Inc. J Polym Sci Part B: Polym Phys 41: 1578–1588, 2003