1. |
|
|
2. |
|
|
3. |
|
|
4. |
- Leal Filho, W., et al.
(author)
-
Handling climate change education at universities : an overview
- 2021
-
In: Environmental Sciences Europe. - : Springer Nature. - 2190-4707 .- 2190-4715. ; 33:1
-
Journal article (peer-reviewed)abstract
- Background: Climate change is a problem which is global in nature, and whose effects go across a wide range of disciplines. It is therefore important that this theme is taken into account as part of universities´ teaching and research programs. Methods: A three-tiered approach was used, consisting of a bibliometric analysis, an online survey and a set of case studies, which allow a profile to be built, as to how a sample of universities from 45 countries handle climate change as part of their teaching programs. Results: This paper reports on a study which aimed at identifying the extent to which matters related to climate change are addressed within the teaching and research practices at universities, with a focus on the training needs of teaching staff. It consists of a bibliometric analysis, combined with an online worldwide survey aimed at ascertaining the degree of involvement from universities in reducing their own carbon footprint, and the ways they offer training provisions on the topic. This is complemented by a set of 12 case studies from universities round the world, illustrating current trends on how universities handle climate change. Apart from reporting on the outcomes of the study, the paper highlights what some universities are doing to handle climate issues, and discusses the implications of the research. Conclusions: The paper lists some items via which universities may better educate and train their students on how to handle the many challenges posed by climate change.
|
|
5. |
|
|
6. |
- Loni, Mohammad, et al.
(author)
-
DeepMaker : A multi-objective optimization framework for deep neural networks in embedded systems
- 2020
-
In: Microprocessors and microsystems. - : Elsevier B.V.. - 0141-9331 .- 1872-9436. ; 73
-
Journal article (peer-reviewed)abstract
- Deep Neural Networks (DNNs) are compute-intensive learning models with growing applicability in a wide range of domains. Due to their computational complexity, DNNs benefit from implementations that utilize custom hardware accelerators to meet performance and response time as well as classification accuracy constraints. In this paper, we propose DeepMaker framework that aims to automatically design a set of highly robust DNN architectures for embedded devices as the closest processing unit to the sensors. DeepMaker explores and prunes the design space to find improved neural architectures. Our proposed framework takes advantage of a multi-objective evolutionary approach that exploits a pruned design space inspired by a dense architecture. DeepMaker considers the accuracy along with the network size factor as two objectives to build a highly optimized network fitting with limited computational resource budgets while delivers an acceptable accuracy level. In comparison with the best result on the CIFAR-10 dataset, a generated network by DeepMaker presents up to a 26.4x compression rate while loses only 4% accuracy. Besides, DeepMaker maps the generated CNN on the programmable commodity devices, including ARM Processor, High-Performance CPU, GPU, and FPGA.
|
|
7. |
- Samperio Ventayol, Pilar, et al.
(author)
-
Bacterial detection by NAIP/NLRC4 elicits prompt contractions of intestinal epithelial cell layers
- 2021
-
In: Proceedings of the National Academy of Sciences of the United States of America. - : Proceedings of the National Academy of Sciences. - 0027-8424 .- 1091-6490. ; 118:16
-
Journal article (peer-reviewed)abstract
- The gut epithelium serves to maximize the surface for nutrient and fluid uptake, but at the same time must provide a tight barrier to pathogens and remove damaged intestinal epithelial cells (IECs) without jeopardizing barrier integrity. How the epithelium coordinates these tasks remains a question of significant interest. We used imaging and an optical flow analysis pipeline to study the dynamicity of untransformed murine and human intestinal epithelia, cultured atop flexible hydrogel supports. Infection with the pathogen Salmonella Typhimurium (S.Tm) within minutes elicited focal contractions with inward movements of up to similar to 1,000 IECs. Genetics approaches and chimeric epithelial monolayers revealed contractions to be triggered by the NAIP/NLRC4 inflammasome, which sensed type-III secretion system and flagellar ligands upon bacterial invasion, converting the local tissue into a contraction epicenter. Execution of the response required swift sublytic Gasdermin D pore formation, ion fluxes, and the propagation of a myosin contraction pulse across the tissue. Importantly, focal contractions preceded, and could be uncoupled from, the death and expulsion of infected IECs. In both two-dimensional monolayers and three-dimensional enteroids, multiple infection-elicited contractions coalesced to produce shrinkage of the epithelium as a whole. Monolayers deficient for Caspase-1(-11) or Gasdermin D failed to elicit focal contractions but were still capable of infected IEC death and expulsion. Strikingly, these monolayers lost their integrity to a markedly higher extent than wild-type counterparts. We propose that prompt NAIP/NLRC4/Caspase-1/Gasdermin D/myosin-dependent contractions allow the epithelium to densify its cell packing in infected regions, thereby preventing tissue disintegration due to the subsequent IEC death and expulsion process.
|
|
8. |
- Sinaei, Sima, et al.
(author)
-
ELC-ECG : Efficient LSTM cell for ECG classification based on quantized architecture
- 2021
-
In: Proceedings - IEEE International Symposium on Circuits and Systems. - : Institute of Electrical and Electronics Engineers Inc.. - 9781728192017 ; May
-
Conference paper (peer-reviewed)abstract
- Long Short-Term Memory (LSTM) is one of the most popular and effective Recurrent Neural Network (RNN) models used for sequence learning in applications such as ECG signal classification. Complex LSTMs could hardly be deployed on resource-limited bio-medical wearable devices due to the huge amount of computations and memory requirements. Binary LSTMs are introduced to cope with this problem. However, naive binarization leads to significant accuracy loss in ECG classification. In this paper, we propose an efficient LSTM cell along with a novel hardware architecture for ECG classification. By deploying 5-level binarized inputs and just 1-level binarization for weights, output, and in-memory cell activations, the delay of one LSTM cell operation is reduced 50x with about 0.004% accuracy loss in comparison with full precision design of ECG classification.
|
|