Ultrathin films of a poly(styrene)-block-poly(2-vinylpyrindine) diblock copolymer (PS-b-P2VP) and poly(styrene)-block-poly(4-vinylpyrindine) diblock copolymer (PS-b-P4VP) were used to form ...surface-induced nanopattern (SINPAT) on mica. Surface interaction controlled microphase separation led to the formation of chemically heterogeneous surface nanopatterns on dry ultrathin films. Two distinct nanopatterned surfaces, namely, wormlike and dotlike patterns, were used to investigate the influence of topography in the nanometer range on cell adhesion, proliferation, and migration. Atomic force microscopy was used to confirm that SINPAT was stable under cell culture conditions. Fibroblasts and mesenchymal progenitor cells were cultured on the nanopatterned surfaces. Phase contrast and confocal laser microscopy showed that fibroblasts and mesenchymal progenitor cells preferred the densely spaced wormlike patterns. Atomic force microscopy showed that the cells remodelled the extracellular matrix differently as they migrate over the two distinctly different nanopatterns.
Ultrathin films of a poly(styrene)-block-poly(2-vinylpyrindine) diblock copolymer (PS-b-P2VP) and poly(styrene)-block-poly(4- vinylpyrindine) diblock copolymer (PS-b-P4VP) were used to form surface- ...induced nanopattern (SINPAT) on mica. Surface interaction controlled microphase separation led to the formation of chemically heterogeneous surface nanopatterns on dry ultrathin films. Two distinct nanopatterned surfaces, namely, wormlike and dotlike patterns, were used to investigate the influence of topography in the nanometer range on cell adhesion, proliferation, and migration. Atomic force microscopy was used to confirm that SINPAT was stable under cell culture conditions. Fibroblasts and mesenchymal progenitor cells were cultured on the nanopatterned surfaces. Phase contrast and confocal laser microscopy showed that fibroblasts and mesenchymal progenitor cells preferred the densely spaced wormlike patterns. Atomic force microscopy showed that the cells remodelled the extracellular matrix differently as they migrate over the two distinctly different nanopatterns.
Model quantization is a widely used technique to compress and accelerate deep neural network (DNN) inference. Emergent DNN hardware accelerators begin to support mixed precision (1-8 bits) to further ...improve the computation efficiency, which raises a great challenge to find the optimal bitwidth for each layer: it requires domain experts to explore the vast design space trading off among accuracy, latency, energy, and model size, which is both time-consuming and sub-optimal. There are plenty of specialized hardware for neural networks, but little research has been done for specialized neural network optimization for a particular hardware architecture. Conventional quantization algorithm ignores the different hardware architectures and quantizes all the layers in a uniform way. In this paper, we introduce the Hardware-Aware Automated Quantization (HAQ) framework which leverages the reinforcement learning to automatically determine the quantization policy, and we take the hardware accelerator's feedback in the design loop. Rather than relying on proxy signals such as FLOPs and model size, we employ a hardware simulator to generate direct feedback signals (latency and energy) to the RL agent. Compared with conventional methods, our framework is fully automated and can specialize the quantization policy for different neural network architectures and hardware architectures. Our framework effectively reduced the latency by 1.4-1.95× and the energy consumption by 1.9× with negligible loss of accuracy compared with the fixed bitwidth (8 bits) quantization. Our framework reveals that the optimal policies on different hardware architectures (i.e., edge and cloud architectures) under different resource constraints (i.e., latency, energy and model size) are drastically different. We interpreted the implication of different quantization policies, which offer insights for both neural network architecture design and hardware architecture design.
We present APQ, a novel design methodology for efficient deep learning deployment. Unlike previous methods that separately optimize the neural network architecture, pruning policy, and quantization ...policy, we design to optimize them in a joint manner. To deal with the larger design space it brings, we devise to train a quantization-aware accuracy predictor that is fed to the evolutionary search to select the best fit. Since directly training such a predictor requires time-consuming quantization data collection, we propose to use predictor-transfer technique to get the quantization-aware predictor: we first generate a large dataset of <NN architecture, ImageNet accuracy> pairs by sampling a pretrained unified supernet and doing direct evaluation; then we use these data to train an accuracy predictor without quantization, further transferring its weights to train the quantization-aware predictor, which largely reduces the quantization data collection time. Extensive experiments on ImageNet show the benefits of this joint design methodology: the model searched by our method maintains the same level accuracy as ResNet34 8-bit model while saving 8x BitOps; we obtain the same level accuracy as MobileNetV2+HAQ while achieving 2×/1.3× latency/energy saving; the marginal search cost of joint optimization for a new deployment scenario outperforms separate optimizations using ProxylessNAS+AMC+HAQ by 2.3% accuracy while reducing 600x GPU hours and CO2 emission.
Depressive disorder is a psychiatric disease characterized by its main symptoms of low mood and anhedonia. Due to its complex etiology, current clinical treatments for depressive disorder are ...limited. In this study, we assessed the role of the δ opioid receptor (δOR) system in the development of chronic-restraint-stressed (CRS)-induced depressive behaviors. We employed a 21-day CRS model and detected the c-fos activation and protein levels' changes in enkephalin (ENK)/δOR. It was found that the hippocampus and amygdala were involved in CRS-induced depression. The expression of pro-enkephalin (PENK), the precursors of the endogenous ligand for δOR, was significantly decreased in the hippocampus and amygdala following CRS. We then treated the mice with SNC80, a specific δOR agonist, to examine its anti-depressant effects in the tail suspension test (TST), forced swimming test (FST), and sucrose preference test (SPT). SNC80 administration significantly reversed depressive-like behaviors, and this antidepressant effect could be blocked by a TrkB inhibitor: ANA-12. Although ANA-12 treatment had no significant effect on the expression of ENK/δOR, it blocked the promoting effects of brain-derived neurotrophic factor (BDNF)/tyrosine kinase B(TrkB) signaling by SNC80 in the hippocampus and amygdala. Therefore, the present study demonstrates that SNC80 exerts anti-depressant effects by up-regulating the BDNF/TrkB signaling pathway in the hippocampus and amygdala in CRS-induced depression and provides evidence that δOR's agonists may be potential anti-depressant therapeutic agents.
•The ENK/δOR system in the hippocampus and amygdala is involved in CRS-induced depression.•The level of ENK and BDNF-TrkB signaling pathway decreased significantly in CRS-induced depression.•The δOR agonist SNC80 could reverse CRS-induced depression behaviors and the decline of BDNF-TrkB signaling pathway levels.
Model quantization is a widely used technique to compress and accelerate deep neural network (DNN) inference. Emergent DNN hardware accelerators begin to support
flexible bitwidth
(1–8 bits) to ...further improve the computation efficiency, which raises a great challenge to find the optimal bitwidth for each layer: it requires domain experts to explore the vast design space trading off accuracy, latency, energy, and model size, which is both time-consuming and usually sub-optimal. There are plenty of specialized hardware accelerators for neural networks, but little research has been done to design specialized neural networks optimized for a particular hardware accelerator. The latter is demanding given the much longer design cycle of silicon than neural nets. Conventional quantization algorithm ignores the different hardware architectures and quantizes all the layers in a uniform way. In this paper, we introduce the Hardware-Aware Automated Quantization (HAQ) framework which automatically determine the quantization policy, and we take the hardware accelerator’s feedback in the design loop. Rather than relying on proxy signals such as FLOPs and model size, we employ a hardware simulator to generate the direct feedback signals to the RL agent. Compared with other conventional methods, our framework is fully automated and can specialize the quantization policy for different neural network architectures and hardware architectures. Our framework effectively reduced the latency by 1.4–1.95
×
and the energy consumption by 1.9
×
with negligible loss of accuracy compared with the fixed bitwidth (8 bits) quantization. Our framework reveals that the optimal policies on different hardware architectures (i.e., edge and cloud architectures) under different resource constraints (i.e., latency, energy and model size) are drastically different. We interpreted the implication of different quantization policies, which offer insights for both neural network architecture design and hardware architecture design.
One of the most frequent side effects of atypical antipsychotics is hyperprolactinemia (HPRL), and metformin or aripiprazole co-prescription is regarded as an effective therapy option for reducing ...prolactin (PRL) levels. However, whether either of the two drugs can reduce PRL levels in patients with long-term hospitalized chronic schizophrenia with co-morbid type 2 diabetes (T2DM) has not been adequately reported.
In our study, long-term hospitalized chronic schizophrenia patients with co-T2DM who were prescribed olanzapine or risperidone as the primary antipsychotic medication were enrolled. A total of 197 of these cases with co-prescribed aripiprazole were set up as the study group (co-Ari group), and the other 204 cases without co-prescribed aripiprazole were set up as the control group (non-Ari group). The two groups' variations in each target parameter were compared, and the variables affecting PRL levels were examined.
Compared to the non-Ari group, fasting blood glucose (FBG), blood uric acid (UA), total cholesterol (TC), triglyceride (TG), and low-density lipoprotein cholesterol (LDL-C) levels were significantly higher in the co-Ari group, but there was no difference in PRL levels. Co-prescribing aripiprazole had no impact on PRL levels in all patients with co-T2DM, and aripiprazole dose had no impact on PRL levels in the clinical subgroup of the co-Ari group.
Aripiprazole not only worsened the severity of index disturbances associated to metabolism in long-term hospitalized chronic schizophrenia patients with co-T2DM on metformin-based hypoglycemic medications but also failed to lower PRL levels.
Early freezes in the spring are becoming increasingly common. To protect farmers from low temperatures (LT), crops need improved LT tolerance. The hypothesis of this paper is that maize (Zea mays L.) ...hybrids released in the 2000s had higher LT tolerances than those released in the 1970s. In the study, eight popular maize released between 1970 and 2010 were subjected to LT in four different trials. Germination time, growth uniformity, recovery ability from LT stress, and leaf productivity at seedling stage are important indicators in evaluating LT tolerance of maize. Hybrids released in the 1970s had a higher gibberellin A3 (GA3) content in seeds, which resulted in faster germination in LT (P = .02). In contrast, plant growth uniformity of the 2000s hybrids were higher than that of the 1970s hybrids in LT (P = .00 at emergence). Upon being returned to warm temperatures, the 2000s hybrids recovered more rapidly than old ones mainly because of the larger root systems, higher GA3, and indole‐3‐acetic acid (IAA) content in seedlings. In addition, the 2000s hybrids had a higher chlorophyll content, Fv/Fm, soluble sugar and starch content, and antioxidant enzyme activity in seedlings than old hybrids in LT, which can also contribute to a fast recovery from LT stress. Based on the growth uniformity and recovery ability from LT stress, new hybrids showed a higher LT tolerance than old hybrids during early growth stage. This study showed the genetic tolerance in LT may exist in hybrids that were released 50 yr ago, and that exporting differences between hybrids possible mechanism of temperature tolerance can be identified.
Core Ideas
New hybrids showed a higher low temperature tolerance than old hybrids during early growth.
Old maize hybrids germinated faster than new ones in low temperature stress.
New maize hybrids grew more uniformly than old ones during LT and recovered faster after low temperature.
Display omitted
•The response of the 7P4R WO3 sensor with a 742-fold enhancement up to 1010 towards 1 ppm H2 at 70 °C.•The sensor can identify ultralow concentration of H2 with a theoretical ...detection limit of 0.252 ppb.•The sensor has good repeatability, selectivity, and high response (650) even at 65 % humidity to 1 ppm H2.•Ex-situ Raman and HER results explained the mechanism of the response’ enhancement.
Hydrogen is the cleanest fuel, but the safe storage and transportation of hydrogen is a relatively troublesome task, thus, developing high-performance hydrogen sensors has certain challenges. In this paper, high-performance hydrogen sensing material at near-ambient temperatures was prepared by using platinum and ruthenium co-modified the surface of WO3 nanowires obtained via electrospinning. When the atomic percentages of platinum (Pt) & ruthenium (Ru) to tungsten (W) is 7:4, the response value of sensors to 1 ppm hydrogen at near room temperature (70 °C) is 1010, making it suitable for practical application. Particularly, the sensor has a low detection limit of 252 ppt, good repeatability, selectivity and long-term stability. In addition, the electrocatalytic dehydrogenation (HER) performance of the samples was investigated and combined with the Raman study to explore the hydrogen sensing mechanism in this work. This work demonstrated that the modification of double noble metals could significantly improve the hydrogen sensing performance of WO3 nanowires, and it is expected to be extended to the practical application of metal oxide semiconductors based chemiresistive hydrogen sensors, which is of great significance for the large-scale safe use of hydrogen energy in the future.
Efficient deep learning inference requires algorithm and hardware codesign to enable specialization: we usually need to change the algorithm to reduce memory footprint and improve energy efficiency. ...However, the extra degree of freedom from the neural architecture design makes the design space much larger: it is not only about designing the hardware architecture but also codesigning the neural architecture to fit the hardware architecture. It is difficult for human engineers to exhaust the design space by heuristics. We propose design automation techniques for architecting efficient neural networks given a target hardware platform. We investigate automatically designing specialized and fast models, auto channel pruning, and auto mixed-precision quantization. We demonstrate that such learning-based, automated design achieves superior performance and efficiency than the rule-based human design. Moreover, we shorten the design cycle by 200× than previous work, so that we can afford to design specialized neural network models for different hardware platforms.
PDF
