For the efficient execution of deep convolutional neural networks (CNN) on edge devices, various approaches have been presented which reduce the bit width of the network parameters down to 1 bit. Binarization of the first layer was always excluded, as it leads to a significant error increase. Here, we present the novel concept of binary input layer (BIL), which allows the usage of binary input data by learning bit specific binary weights. The concept is evaluated on three datasets (PAMAP2, SVHN, CIFAR-10). Our results show that this approach is in particular beneficial for multimodal datasets (PAMAP2) where it outperforms networks using full precision weights in the first layer by 1.92 percentage points (pp) while consuming only 2 % of the chip area.32nd Conference on Neural Information Processing Systems (NIPS 2018),
An electrochemical platform for generating and controlling
a localized
pH microenvironment on demand is proposed by employing a closed-loop
control algorithm based on an iridium oxide pH sensor input. We use
a combination of solution-borne quinones and galvanostatic excitation
on a prepatterned indium tin oxide (ITO) working electrode to modulate
pH within a very well confined, small volume of solution close to
the electrode surface. We demonstrate that the rate of pH change can
be controlled at up to 2 pH s–1 with an excellent
repeatability (±0.004). The desired pH microenvironment can be
stably maintained for longer than 2 h within ±0.0012 pH. As a
high-impact application of the platform technology, we propose a single-step
immunoassay and demonstrate its utility in measuring C-reactive protein
(CRP), a critical inflammatory marker in various conditions such as
myocardial infarction and even SARS-Cov-2. Utilizing pH modulation
technology along with pH-sensitive fluorescence dye simplifies the
immunoassay process into a single-step, where a mixture of all of
the reagents is incubated only for 1 h without any washing steps or
the need to change solution. This simplified immunoassay process minimizes
the hands-on time of the end-user and thus decreases technician-driven
errors. Moreover, the absence of complicated liquid-handling hardware
makes it more suitable and attractive for an ultracompact platform
to ultimately be used in a point-of-care diagnostic assay.
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.