Convolutional neural network (CNN) has become a successful algorithm in the region of artificial intelligence and a strong candidate for many computer vision algorithms. But the computation ...complexity of CNN is much higher than traditional algorithms. With the help of GPU acceleration, CNN-based applications are widely deployed in servers. However, for embedded platforms, CNN-based solutions are still too complex to be applied. Various dedicated hardware designs on field-programmable gate arrays (FPGAs) have been carried out to accelerate CNNs, while few of them explore the whole design flow for both fast deployment and high power efficiency. In this paper, we investigate state-of-the-art CNN models and CNN-based applications. Requirements on memory, computation and the flexibility of the system are summarized for mapping CNN on embedded FPGAs. Based on these requirements, we propose Angel-Eye, a programmable and flexible CNN accelerator architecture, together with data quantization strategy and compilation tool. Data quantization strategy helps reduce the bit-width down to 8-bit with negligible accuracy loss. The compilation tool maps a certain CNN model efficiently onto hardware. Evaluated on Zynq XC7Z045 platform, Angel-Eye is 6× faster and 5× better in power efficiency than peer FPGA implementation on the same platform. Applications of VGG network, pedestrian detection and face alignment are used to evaluate our design on Zynq XC7Z020. NIVIDA TK1 and TX1 platforms are used for comparison. Angel-Eye achieves similar performance and delivers up to 16× better energy efficiency.
The convolutional neural network (CNN) has become a state-of-the-art method for several artificial intelligence domains in recent years. The increasingly complex CNN models are both computation-bound ...and I/O-bound. Field-programmable gate array-based accelerators driven by custom instruction set architecture (ISA) achieve a balance between generality and efficiency, but there is much on them left to be optimized. We propose the full-stack compiler deep neural network virtual machine (DNNVM), which is an integration of optimizers for graphs, loops and data layouts, an assembler, a runtime supporter, and a validation environment. The DNNVM works in the context of deep learning frameworks and transforms CNN models into the directed acyclic graph: XGraph. Based on XGraph, we transform the optimization challenges for both data layout and pipeline into graph-level problems. DNNVM enumerates all potentially profitable fusion opportunities by a heuristic subgraph isomorphism algorithm to leverage pipeline and data layout optimizations, and searches for the best choice of execution strategies of the whole computing graph. On the Xilinx ZU2@330 MHz and ZU9@330 MHz, we achieve equivalently state-of-the-art performance on our benchmarks by naïve implementations without optimizations, and the throughput is further improved up to <inline-formula> <tex-math notation="LaTeX">1.26\times </tex-math></inline-formula> by leveraging heterogeneous optimizations in DNNVM. Finally, with ZU9@330 MHz, we achieve state-of-the-art performance for VGG and ResNet50. We achieve a throughput of 2.82 TOPs/s and an energy efficiency of 123.7 GOPs/s/W for VGG. Additionally, we achieve 1.38 TOPs/s for ResNet50 and 1.41 TOPs/s for GoogleNet.
The convolutional neural network (CNN) has become a state-of-the-art method for several artificial intelligence domains in recent years. The increasingly complex CNN models are both computation-bound ...and I/O-bound. FPGA-based accelerators driven by custom instruction set architecture (ISA) achieve a balance between generality and efficiency, but there is much on them left to be optimized. We propose the full-stack compiler DNNVM, which is an integration of optimizers for graphs, loops and data layouts, and an assembler, a runtime supporter and a validation environment. The DNNVM works in the context of deep learning frameworks and transforms CNN models into the directed acyclic graph: XGraph. Based on XGraph, we transform the optimization challenges for both the data layout and pipeline into graph-level problems. DNNVM enumerates all potentially profitable fusion opportunities by a heuristic subgraph isomorphism algorithm to leverage pipeline and data layout optimizations, and searches for the best choice of execution strategies of the whole computing graph. On the Xilinx ZU2 @330 MHz and ZU9 @330 MHz, we achieve equivalently state-of-the-art performance on our benchmarks by na\"ive implementations without optimizations, and the throughput is further improved up to 1.26x by leveraging heterogeneous optimizations in DNNVM. Finally, with ZU9 @330 MHz, we achieve state-of-the-art performance for VGG and ResNet50. We achieve a throughput of 2.82 TOPs/s and an energy efficiency of 123.7 GOPs/s/W for VGG. Additionally, we achieve 1.38 TOPs/s for ResNet50 and 1.41 TOPs/s for GoogleNet.
To investigate the effect of aerobic exercise on five dimensions of physical self-worth, exercise capacity, physical condition, physical attractiveness, and physical quality in body self-esteem of ...Chinese college students. By searching PubMed, Web of Science, Cochrane Library, CNIK database, VIP database, WANFANG database platform, we searched for the subject terms or keywords "body self-esteem", "Chinese college students", "Systematic evaluation", "Aerobic exercise", "Exercise intervention", "Meta-Analysis". The search method was a combination of subject terms and keywords and title, and the search period was from database creation to The search was conducted from database creation to May 2022. A total of 3221 articles were searched, and 9 articles were included in the study through repeated screening. Risk of bias was assessed with Cochrane and the quality of studies in the literature was assessed using Grade pro software. The outcome indicators of the included literature were analysed using review manager 5.4 software and StataMP 17.0 software. Nine papers including 1613 subjects were included. results of Meta-analysis showed that aerobic exercise was effective in improving physical self-worth (WMD = 1.46, 95% CI: 1.08-1.83, p<0.001), improving exercise capacity (WMD = 1.62, 95% CI: 1.23-2.01, p<0.001), improving physical attractiveness (WMD = 1.32, 95% CI: 0.98-1.67, p<0.001), improved physical condition (WMD = 1.32, 95% CI: 0.98-1.67, P<0.001), improved physical fitness (WMD = 1.51, 95% CI: 1.07-1.95, P<0.001). The differences were all statistically significant. Aerobic exercise can effectively improve the body self-esteem of Chinese college students. In exercise, male students pursue is athletic ability and physical fitness, and female students pursue is the sense of physical self-worth and physical attractiveness. Aerobic exercise has a greater increase in body self-esteem for obese or Obese college students. Aerobics and physical dance are the most cost-effective for improving body self-esteem. Medium-intensity relative to low-intensity exercise was effective for body self-esteem intervention. A single exercise session of 90 minutes was more effective than a single 30-minute session in boosting body self-esteem, and the overall intervention duration of 16 weeks was more effective than 10 weeks.
Cloud microphysical properties and precipitation over the Tibetan Plateau are unique because of the high terrains, clean atmosphere, and sufficient water vapor. With dual‐polarization precipitation ...radar and cloud radar measurements during the Third Tibetan Plateau Atmospheric Scientific Experiment, the simulated microphysics and precipitation by the Weather Research and Forecasting (WRF) model with the Chinese Academy of Meteorological Sciences (CAMS) microphysics and other microphysical schemes are investigated through a typical plateau rainfall event on 22 July 2014. Results show that the WRF‐CAMS simulation reasonably reproduces the spatial distribution of 24 h accumulated precipitation but has limitations in simulating time evolution of precipitation rates. The model‐calculated polarimetric radar variables have biases as well, suggesting bias in modeled hydrometeor types. The raindrop sizes in convective region are larger than those in stratiform region indicated by the small intercept of raindrop size distribution in the former. In addition, the warm rain processes generate heavier precipitation than the cold rain processes do over the rainfall centers during weak convection period. The sensitivity of precipitation to perturbing the warm rain microphysical processes show that doubling droplet condensation increases precipitation significantly and produces the best area‐averaged rain rate, suggesting biases in thermodynamics in the baseline simulation. Halving raindrop evaporation results in an increase in weak rainfall areas along with a warmer subcloud layer. Increasing the initial cloud droplet size causes the rain rate reduced by half, an opposite effect to that of increasing droplet condensation.
Key Points
Warm rain is important for TP weak convective precipitation over the rainfall centers
Precipitation is sensitive to cloud droplet condensation in the weak convection period
Raindrop sizes in convective region are larger than those in stratiform region
BackgroundNon-small cell lung cancer (NSCLC) patients bearing targetable oncogene alterations typically derive limited benefit from immune checkpoint blockade (ICB), which has been attributed to low ...tumor mutation burden (TMB) and/or PD-L1 levels. We investigated oncogene-specific differences in these markers and clinical outcome.MethodsThree cohorts of NSCLC patients with oncogene alterations (n=4189 total) were analyzed. Two clinical cohorts of advanced NSCLC patients treated with ICB monotherapy MD Anderson (MDACC; n=172) and Flatiron Health-Foundation Medicine Clinico-Genomic Database (CGDB; n=894 patients) were analyzed for clinical outcome. The FMI biomarker cohort (n=4017) was used to assess the association of oncogene alterations with TMB and PD-L1 expression.ResultsHigh PD-L1 expression (PD-L1 ≥50%) rate was 19%–20% in classic EGFR, EGFR exon 20 and HER2-mutant tumors, and 34%–55% in tumors with ALK, BRAF V600E, ROS1, RET, or MET alterations. Compared with KRAS-mutant tumors, BRAF non-V600E group had higher TMB (9.6 vs KRAS 7.8 mutations/Mb, p=0.003), while all other oncogene groups had lower TMB (p<0.001). In the two clinical cohorts treated with ICB, molecular groups with EGFR, HER2, ALK, ROS1, RET, or MET alterations had short progression-free survival (PFS; 1.8–3.7 months), while BRAF V600E group was associated with greater clinical benefit from ICB (CGDB cohort: PFS 9.8 months vs KRAS 3.7 months, HR 0.66, p=0.099; MDACC cohort: response rate 62% vs KRAS 24%; PFS 7.4 vs KRAS 2.8 months, HR 0.36, p=0.026). KRAS G12C and non-G12C subgroups had similar clinical benefit from ICB in both cohorts. In a multivariable analysis, BRAF V600E mutation (HR 0.58, p=0.041), PD-L1 expression (HR 0.57, p=0.022), and high TMB (HR 0.66, p<0.001) were associated with longer PFS.ConclusionsHigh TMB and PD-L1 expression are predictive for benefit from ICB treatment in oncogene-driven NSCLCs. NSCLC harboring BRAF mutations demonstrated superior benefit from ICB that may be attributed to higher TMB and higher PD-L1 expression in these tumors. Meanwhile EGFR and HER2 mutations and ALK, ROS1, RET, and MET fusions define NSCLC subsets with minimal benefit from ICB despite high PD-L1 expression in NSCLC harboring oncogene fusions. These findings indicate a TMB/PD-L1-independent impact on sensitivity to ICB for certain oncogene alterations.
Display omitted
•98.84% of titanium is extracted by NH4HF2-HF leaching process.•The MgTi2O5 and Al2TiO5 in MxTi3-xO5 (0<x<2, M=Ti, Mg, Al, etc) are converted to TiF62− and Mg-Al-bearing precipitate, ...respectively.•The TiF62− is converted to (NH4)2TiOF4.•The final titanium dioxide product containing 99.88wt% TiO2 is obtained.•The well crystallized anatase and rutile are obtained through roasting the titanium dioxide at 800°C and 1000°C, respectively.
A novel process to prepare titanium dioxide from Ti-bearing electric furnace slag by NH4HF2-HF leaching and hydrolyzing process has been developed. In this present study, the effects of NH4+/F mXolar ratio, leaching temperature, F concentration, liquid/solid mass ratio, leaching time on the Ti extraction, and the phase transformations have been investigated to reveal the leaching mechanism of Ti-bearing electric furnace slag in NH4HF2-HF solution. In the NH4HF2-HF leaching process, the MgTi2O5 and Al2TiO5 are converted to TiF62− and Mg-Al-bearing precipitate. Ti extraction rate reached 98.84% under the optimal conditions. In addition, 98.25% iron ions can be removed in the presence of NaCl prior to hydrolysis process. The effects of pH and temperature on the selective hydrolysis of TiF62− during hydrolysis process were also studied. In the hydrolysis process, the TiF62− is converted to (NH4)2TiOF4. By calcination, high grade TiO2 powder with its purity of 99.88% was obtained, using which the products, well crystallized anatase and rutile, were obtained through roasting at 800°C and 1000°C, respectively.
In the third-harmonic-generation crystal of Inertial Confinement Fusion (ICF) system, the transverse stimulated Raman scattering (TSRS) is one of the most destructive nonlinear scattering effects. In ...this paper, the effects of time-phase modulation on TSRS are studied in KDP and KD*P. The time-phase modulation is analyzed theoretically and “the fastest growth condition of Stokes light” is proposed. And the finite difference method is used to calculate Maxwell-Bloch-Langevin equations describing TSRS. The results show that the time-phase modulation can greatly reduce peak intensity and peak fluence of Stokes light. In addition, the effects of time-phase modulation on TSRS are also analyzed in detail when time-phase modulation model, pump pulse width, pump peak intensity and pump beam aperture are changed. By analysis and calculation, we give reasonable suggestions for the selection of the above working parameters.
•In this paper, the effects of time-phase modulation on TSRS are studied in KDP and KD*P.•The time-phase modulation is analyzed theoretically and “the fastest growth condition of Stokes light” is proposed.•And the finite difference method is used to calculate Maxwell-Bloch-Langevin equations describing TSRS.
On the origin of helical mesostructures Yang, Sui; Zhao, Lingzhi; Yu, Chengzhong ...
Journal of the American Chemical Society,
08/2006, Letnik:
128, Številka:
32
Journal Article
Recenzirano
The investigation on the formation mechanism of helical structures and the synthesis of helical materials is attractive for scientists in different fields. Here we report the synthesis of helical ...mesoporous materials with chiral channels in the presence of achiral surfactants. More importantly, we suggest a simple and purely interfacial interaction mechanism to explain the spontaneous formation of helical mesostructures. Unlike the proposed model for the formation of helical molecular chains or surpramolecular packing based on the geometrically motivated model or the entropically driven model, the origin of the helical mesostructured materials may be attributed to a morphological transformation accompanied by a reduction in surface free energy. After the helical morphology is formed, the increase in bending energy together with the derivation from a perfect hexagonal mesostructure may limit the curvature of helices. Our model may be general and important in the designed synthesis of helical mesoporous materials.
Polarization is an important dimension in the research and applications of light waves. However, traditional polarization optics often only focus on the polarization characteristics in the transverse ...plane. Here, we demonstrate a new scheme for the generation of longitudinally varied polarization state in terahertz beam using all-silicon metasurface. We employ wavefront transformation designs with long-focal-depth for orthogonal circularly polarized terahertz waves, achieving varied amplitude and phase along the propagation direction in opposite spin states. Based on the principle of coherent superposition of polarized waves, different linear and elliptical polarization states are obtained in transverse planes along the propagation path, with variable ellipticity and azimuth angle. Simulation results show that a large-scale evolution of the elliptical polarization azimuth angle from 45° to -60° and ellipticity from 20° to -74° can be observed within a focal depth range of 0.45-0.8 mm. We also intuitively display the helical trajectory of the polarization state from left-hand elliptical ones to right-hand elliptical ones within the focal depth range, using the Poincaré sphere. This work expands the application of metasurface devices for multifunctional polarization devices and can be applied to polarization generation and transformation for optical imaging or terahertz communications.