| 1. |
- Chen, Qinyu, et al.
(författare)
-
An Efficient Streaming Accelerator for Low Bit-Width Convolutional Neural Networks
- 2019
-
Ingår i: Electronics. - : MDPI. - 2079-9292. ; 8:4
-
Tidskriftsartikel (refereegranskat)abstract
- Convolutional Neural Networks (CNNs) have been widely applied in various fields, such as image recognition, speech processing, as well as in many big-data analysis tasks. However, their large size and intensive computation hinder their deployment in hardware, especially on the embedded systems with stringent latency, power, and area requirements. To address this issue, low bit-width CNNs are proposed as a highly competitive candidate. In this paper, we propose an efficient, scalable accelerator for low bit-width CNNs based on a parallel streaming architecture. With a novel coarse grain task partitioning (CGTP) strategy, the proposed accelerator with heterogeneous computing units, supporting multi-pattern dataflows, can nearly double the throughput for various CNN models on average. Besides, a hardware-friendly algorithm is proposed to simplify the activation and quantification process, which can reduce the power dissipation and area overhead. Based on the optimized algorithm, an efficient reconfigurable three-stage activation-quantification-pooling (AQP) unit with the low power staged blocking strategy is developed, which can process activation, quantification, and max-pooling operations simultaneously. Moreover, an interleaving memory scheduling scheme is proposed to well support the streaming architecture. The accelerator is implemented with TSMC 40 nm technology with a core size of . It can achieve TOPS/W energy efficiency and area efficiency at 100.1mW, which makes it a promising design for the embedded devices.
|
|
| 2. |
- Chen, Qinyu, et al.
(författare)
-
Smilodon : An Efficient Accelerator for Low Bit-Width CNNs with Task Partitioning
- 2019
-
Ingår i: 2019 IEEE INTERNATIONAL SYMPOSIUM ON CIRCUITS AND SYSTEMS (ISCAS). - : IEEE. - 9781728103976
-
Konferensbidrag (refereegranskat)abstract
- Convolutional Neural Networks (CNNs) have been widely applied in various fields such as image and video recognition, recommender systems, and natural language processing. However, the massive size and intensive computation loads prevent its feasible deployment in practice, especially on the embedded systems. As a highly competitive candidate, low bit-width CNNs are proposed to enable efficient implementation. In this paper, we propose Smilodon, a scalable, efficient accelerator for low bit-width CNNs based on a parallel streaming architecture, optimized with a task partitioning strategy. We also present the 3D systolic-like computing arrays fitting for convolutional layers. Our design is implemented on Zynq XC7ZO20 FPGA, which can satisfy the needs of real-time with a frame rate of 1, 622 FPS throughput, while consuming 2.1 Watt. To the best of our knowledge, our accelerator is superior to the state-of-the-art works in the tradeoff among throughput, power efficiency, and area efficiency.
|
|
| 3. |
- Jiang, Zhiyong, et al.
(författare)
-
Gene flow and an anomaly zone complicate phylogenomic inference in a rapidly radiated avian family (Prunellidae)
- 2024
-
Ingår i: BMC Biology. - : BioMed Central (BMC). - 1741-7007. ; 22:1
-
Tidskriftsartikel (refereegranskat)abstract
- BackgroundResolving the phylogeny of rapidly radiating lineages presents a challenge when building the Tree of Life. An Old World avian family Prunellidae (Accentors) comprises twelve species that rapidly diversified at the Pliocene-Pleistocene boundary.ResultsHere we investigate the phylogenetic relationships of all species of Prunellidae using a chromosome-level de novo assembly of Prunella strophiata and 36 high-coverage resequenced genomes. We use homologous alignments of thousands of exonic and intronic loci to build the coalescent and concatenated phylogenies and recover four different species trees. Topology tests show a large degree of gene tree-species tree discordance but only 40-54% of intronic gene trees and 36-75% of exonic genic trees can be explained by incomplete lineage sorting and gene tree estimation errors. Estimated branch lengths for three successive internal branches in the inferred species trees suggest the existence of an empirical anomaly zone. The most common topology recovered for species in this anomaly zone was not similar to any coalescent or concatenated inference phylogenies, suggesting presence of anomalous gene trees. However, this interpretation is complicated by the presence of gene flow because extensive introgression was detected among these species. When exploring tree topology distributions, introgression, and regional variation in recombination rate, we find that many autosomal regions contain signatures of introgression and thus may mislead phylogenetic inference. Conversely, the phylogenetic signal is concentrated to regions with low-recombination rate, such as the Z chromosome, which are also more resistant to interspecific introgression.ConclusionsCollectively, our results suggest that phylogenomic inference should consider the underlying genomic architecture to maximize the consistency of phylogenomic signal.
|
|
| 4. |
- Meng, Weida, et al.
(författare)
-
Genotype-dependent epigenetic regulation of DLGAP2 in alcohol use and dependence
- 2021
-
Ingår i: Molecular Psychiatry. - : Springer Nature. - 1359-4184 .- 1476-5578. ; 26:8, s. 4367-4382
-
Tidskriftsartikel (refereegranskat)abstract
- Alcohol misuse is a major public health problem originating from genetic and environmental risk factors. Alterations in the brain epigenome may orchestrate changes in gene expression that lead to alcohol misuse and dependence. Through epigenome-wide association analysis of DNA methylation from human brain tissues, we identified a differentially methylated region, DMR-DLGAP2, associated with alcohol dependence. Methylation within DMR-DLGAP2 was found to be genotype-dependent, allele-specific and associated with reward processing in brain. Methylation at the DMR-DLGAP2 regulated expression of DLGAP2 in vitro, and Dlgap2-deficient mice showed reduced alcohol consumption compared with wild-type controls. These results suggest that DLGAP2 may be an interface for genetic and epigenetic factors controlling alcohol use and dependence.
|
|
| 5. |
- Shu, Rui, et al.
(författare)
-
Solid-State Janus Nanoprecipitation Enables Amorphous-Like Heat Conduction in Crystalline Mg3Sb2-Based Thermoelectric Materials
- 2022
-
Ingår i: Advanced Science. - : Wiley. - 2198-3844. ; 9:25
-
Tidskriftsartikel (refereegranskat)abstract
- Solid-state precipitation can be used to tailor material properties, ranging from ferromagnets and catalysts to mechanical strengthening and energy storage. Thermoelectric properties can be modified by precipitation to enhance phonon scattering while retaining charge-carrier transmission. Here, unconventional Janus-type nanoprecipitates are uncovered in Mg3Sb1.5Bi0.5 formed by side-by-side Bi- and Ge-rich appendages, in contrast to separate nanoprecipitate formation. These Janus nanoprecipitates result from local comelting of Bi and Ge during sintering, enabling an amorphous-like lattice thermal conductivity. A precipitate size effect on phonon scattering is observed due to the balance between alloy-disorder and nanoprecipitate scattering. The thermoelectric figure-of-merit ZT reaches 0.6 near room temperature and 1.6 at 773 K. The Janus nanoprecipitation can be introduced into other materials and may act as a general property-tailoring mechanism.
|
|
| 6. |
- Terenius, Olle, et al.
(författare)
-
RNA interference in Lepidoptera : An overview of successful and unsuccessful studies and implications for experimental design
- 2011
-
Ingår i: Journal of insect physiology. - : Elsevier BV. - 0022-1910 .- 1879-1611. ; 57:2, s. 231-245
-
Tidskriftsartikel (refereegranskat)abstract
- Gene silencing through RNA interference (RNAi) has revolutionized the study of gene function, particularly in non-model insects. However, in Lepidoptera (moths and butterflies) RNAi has many times proven to be difficult to achieve. Most of the negative results have been anecdotal and the positive experiments have not been collected in such a way that they are possible to analyze. In this review, we have collected detailed data from more than 150 experiments including all to date published and many unpublished experiments. Despite a large variation in the data, trends that are found are that RNAi is particularly successful in the family Saturniidae and in genes involved in immunity. On the contrary, gene expression in epidermal tissues seems to be most difficult to silence. In addition, gene silencing by feeding dsRNA requires high concentrations for success. Possible causes for the variability of success in RNAi experiments in Lepidoptera are discussed. The review also points to a need to further investigate the mechanism of RNAi in lepidopteran insects and its possible connection to the innate immune response. Our general understanding of RNAi in Lepidoptera will be further aided in the future as our public database at http://insectacentral.org/RNAi will continue to gather information on RNAi experiments.
|
|
| 7. |
- Turchi, Patrice E. A., et al.
(författare)
-
Interface between quantum-mechanical-based approaches, experiments, and CALPHAD methodology
- 2007
-
Ingår i: Calphad. - : Elsevier BV. - 0364-5916 .- 1873-2984. ; 31:1, s. 4-27
-
Tidskriftsartikel (refereegranskat)abstract
- The increased application of quantum-mechanical-based methodologies to the study of alloy stability has required a re-assessment of the field. The focus is mainly on inorganic materials in the solid state. In a first part, after a brief overview of the so-called ab initio methods with their approximations, constraints, and limitations, recommendations are made for a good usage of first-principles codes with a set of qualifiers. Examples are given to illustrate the power and the limitations of ab initio codes. However, despite the "success" of these methodologies, thermodynamics of complex multi-component alloys, as used in engineering applications, requires a more versatile approach presently afforded within CALPHAD. Hence, in a second part, the links that presently exist between ab initio methodologies, experiments, and the CALPHAD approach are examined with illustrations. Finally, the issues of dynamical instability and of the role of lattice vibrations that still constitute the subject of ample discussions within the CALPHAD community are revisited in the light of our current knowledge with a set of recommendations.
|
|
| 8. |
|
|
| 9. |
- Zhu, Changlian, 1964, et al.
(författare)
-
Erythropoietin improved neurologic outcomes in newborns with hypoxic-ischemic encephalopathy.
- 2009
-
Ingår i: Pediatrics. - : American Academy of Pediatrics (AAP). - 1098-4275 .- 0031-4005. ; 124:2
-
Tidskriftsartikel (refereegranskat)abstract
- OBJECTIVE: The purpose of this study was to evaluate the efficacy and safety of erythropoietin in neonatal hypoxic-ischemic encephalopathy (HIE), by using a randomized, prospective study design. METHODS: A total of 167 term infants with moderate/severe HIE were assigned randomly to receive either erythropoietin (N = 83) or conventional treatment (N = 84). Recombinant human erythropoietin, at either 300 U/kg (N = 52) or 500 U/kg (N = 31), was administered every other day for 2 weeks, starting <48 hours after birth. The primary outcome was death or disability. Neurodevelopmental outcomes were assessed at 18 months of age. RESULTS: Complete outcome data were available for 153 infants. Nine patients dropped out during treatment, and 5 patients were lost to follow-up monitoring. Death or moderate/severe disability occurred for 35 (43.8%) of 80 infants in the control group and 18 (24.6%) of 73 infants in the erythropoietin group (P = .017) at 18 months. The primary outcomes were not different between the 2 erythropoietin doses. Subgroup analyses indicated that erythropoietin improved long-term outcomes only for infants with moderate HIE (P = .001) and not those with severe HIE (P = .227). No negative hematopoietic side effects were observed. CONCLUSION: Repeated, low-dose, recombinant human erythropoietin treatment reduced the risk of disability for infants with moderate HIE, without apparent side effects.
|
|
