State-of-the-art deep neural networks (DNNs) have hundreds of millions of connections and are both computationally and memory intensive, making them difficult to deploy on embedded systems with limited hardware resources and power budgets. While custom hardware helps the computation, fetching weights from DRAM is two orders of magnitude more expensive than ALU operations, and dominates the required power.Previously proposed 'Deep Compression' makes it possible to fit large DNNs (AlexNet and VGGNet) fully in on-chip SRAM. This compression is achieved by pruning the redundant connections and having multiple connections share the same weight. We propose an energy efficient inference engine (EIE) that performs inference on this compressed network model and accelerates the resulting sparse matrix-vector multiplication with weight sharing. Going from DRAM to SRAM gives EIE 120× energy saving; Exploiting sparsity saves 10×; Weight sharing gives 8×; Skipping zero activations from ReLU saves another 3×. Evaluated on nine DNN benchmarks, EIE is 189× and 13× faster when compared to CPU and GPU implementations of the same DNN without compression. EIE has a processing power of 102 GOPS/s working directly on a compressed network, corresponding to 3 TOPS/s on an uncompressed network, and processes FC layers of AlexNet at 1.88×10 4 frames/sec with a power dissipation of only 600mW. It is 24,000× and 3,400× more energy efficient than a CPU and GPU respectively. Compared with DaDianNao, EIE has 2.9×, 19× and 3× better throughput, energy efficiency and area efficiency.
The strategies needed to promote equity in the cardiovascular health of African Americans require input from a broad set of stakeholders, including clinicians and researchers from across multiple disciplines.
There is increasing interest in using nanopores in synthetic membranes as resistive-pulse sensors for biomedical analytes. Analytes detected with prototype artificial-nanopore biosensors include drugs, DNA, proteins, and viruses. This field is, however, currently in its infancy. A key question that must be addressed in order for such sensors to progress from an interesting laboratory experiment to practical devices is: Can the artificial-nanopore sensing element be reproducibly prepared? We have been evaluating sensors that employ a conically shaped nanopore prepared by the track-etch method as the sensor element. We describe here a new two-step pore-etching procedure that allows for good reproducibility in nanopore fabrication. In addition, we describe a simple mathematical model that allows us to predict the characteristics of the pore produced given the experimental parameters of the two-step etch. This method and model constitute important steps toward developing practical, real-world, artificial-nanopore biosensors.
We demonstrate here a new electrokinetic phenomenon, Electroosmotic flow (EOF) rectification, in synthetic membranes containing asymmetric pores. Mica membranes with pyramidally shaped pores prepared by the track-etch method were used. EOF was driven through these membranes by using an electrode in solutions on either side to pass a constant ionic current through the pores. The velocity of EOF depends on the polarity of the current. A high EOF velocity is obtained when the polarity is such that EOF is driven from the larger base opening to the smaller tip opening of the pore. A smaller EOF velocity is obtained when the polarity is reversed such that EOF goes from tip to base. We show that this rectified EOF phenomenon is the result of ion current-rectification observed in such asymmetric-pore membranes.
Overweight/obesity, advanced maternal age, family history of type 2 diabetes, and foreign-borne status are important risk factors for GDM. The relative contributions of these risk factors differ by race/ethnicity, mainly due to differences in population prevalence of these risk factors.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.