Domain Generalization by Solving Jigsaw Puzzles Carlucci, Fabio M.; D'Innocente, Antonio; Bucci, Silvia ...
2019 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR),
06/2019
Conference Proceeding
Odprti dostop
Human adaptability relies crucially on the ability to learn and merge knowledge both from supervised and unsupervised learning: the parents point out few important concepts, but then the children ...fill in the gaps on their own. This is particularly effective, because supervised learning can never be exhaustive and thus learning autonomously allows to discover invariances and regularities that help to generalize. In this paper we propose to apply a similar approach to the task of object recognition across domains: our model learns the semantic labels in a supervised fashion, and broadens its understanding of the data by learning from self-supervised signals how to solve a jigsaw puzzle on the same images. This secondary task helps the network to learn the concepts of spatial correlation while acting as a regularizer for the classification task. Multiple experiments on the PACS, VLCS, Office-Home and digits datasets confirm our intuition and show that this simple method outperforms previous domain generalization and adaptation solutions. An ablation study further illustrates the inner workings of our approach.
Glycerol is a valuable by-product in the biodiesel industries. However, the increase in biodiesel production resulted in an excess production of glycerol, with a limited market compared to its ...availability. Precisely because glycerol became a waste to be disposed of, the costs of biodiesel production have reduced. From an environmental point of view, identifying reactions that can convert glycerol into new products that can be reused in different applications has become a real necessity. According to the unique structural characteristics of glycerol, transformation processes can lead to different chemical functionalities through redox reactions, dehydration, esterification, and etherification, with the formation of products that can be applied both at the finest chemical level and to bulk chemistry.
The selection of a thermal comfort model for establishing indoor optimal hygrothermal conditions during the hot period has a major impact on energy consumption of Net Zero Energy Buildings in hot ...climates. The objective of this paper is to compare the influence of using different thermal comfort models for zero energy buildings in hot climates. The paper compares the impact of applying Fanger's model, Givoni's model, the ASHRAE 55 adaptive comfort model and the EN 15251 adaptive comfort model on energy consumption and comfort performance. Using both the building performance simulation tools ZEBO and EnergyPlus for energy simulation, an existing prototype of a residential apartment module is used to evaluate energy performance and thermal comfort in two parametric series. The first one is the result of coupling natural ventilation and mechanical cooling and the second one is guided coupling natural ventilation, mechanical cooling and ceiling fans. This study shows that the percentage of energy consumption difference meeting the comfort criteria according to ISO 7730 in comparison to EN 15251, ASHRAE 55 or Givoni's model varied up to 16%, 21% and 24.7%, respectively for the presented case study. More energy savings can be expected for buildings in hot climates with greater cooling demands.
Self-Supervised Learning Across Domains Bucci, Silvia; D'Innocente, Antonio; Liao, Yujun ...
IEEE transactions on pattern analysis and machine intelligence,
09/2022, Letnik:
44, Številka:
9
Journal Article
Recenzirano
Odprti dostop
Human adaptability relies crucially on learning and merging knowledge from both supervised and unsupervised tasks: the parents point out few important concepts, but then the children fill in the gaps ...on their own. This is particularly effective, because supervised learning can never be exhaustive and thus learning autonomously allows to discover invariances and regularities that help to generalize. In this paper we propose to apply a similar approach to the problem of object recognition across domains: our model learns the semantic labels in a supervised fashion, and broadens its understanding of the data by learning from self-supervised signals on the same images. This secondary task helps the network to focus on object shapes, learning concepts like spatial orientation and part correlation, while acting as a regularizer for the classification task over multiple visual domains. Extensive experiments confirm our intuition and show that our multi-task method, combining supervised and self-supervised knowledge, provides competitive results with respect to more complex domain generalization and adaptation solutions. It also proves its potential in the novel and challenging predictive and partial domain adaptation scenarios.
The depletion of fossil fuels, attributable to the rapid increase in the world’s population and the growth of industrialization, is estimated to run out in less than ten decades if not replaced by ...alternative energy sources ...
•Using nano-diesel-biodiesel increased the brake power, torque and decrease bsfc of CI engine.•The CO2 and NOx increased while the concentration of CO and HC were decreased with nano-biodiesel-diesel ...blends.•Good correlation was observed between genetic programming predicted results and experimental data.•GP proved to be a useful tool for correlation and simulation of engine parameters.•GP provided an accurate and simple approach in the analysis of the CI engine performance and emissions.
The performance and the exhaust emissions of a diesel engine operating on nano-diesel-biodiesel blended fuels has been investigated. Multi wall carbon nano tubes (CNT) (40, 80 and 120ppm) and nano silver particles (40, 80 and 120ppm) were produced and added as additive to the biodiesel-diesel blended fuel. Six cylinders, four-stroke diesel engine was fuelled with these new blended fuels and operated at different engine speeds. Experimental test results indicated the fact that adding nano particles to diesel and biodiesel fuels, increased diesel engine performance variables including engine power and torque output up to 2% and brake specific fuel consumption (bsfc) was decreased 7.08% compared to the net diesel fuel. CO2 emission increased maximum 17.03% and CO emission in a biodiesel-diesel fuel with nano-particles was lower significantly (25.17%) compared to pure diesel fuel. UHC emission with silver nano-diesel-biodiesel blended fuel decreased (28.56%) while with fuels that contains CNT nano particles increased maximum 14.21%. With adding nano particles to the blended fuels, NOx increased 25.32% compared to the net diesel fuel. This study also presents genetic programming (GP) based model to predict the performance and emission parameters of a CI engine in terms of nano-fuels and engine speed. Experimental studies were completed to obtain training and testing data. The optimum models were selected according to statistical criteria of root mean square error (RMSE) and coefficient of determination (R2). It was observed that the GP model can predict engine performance and emission parameters with correlation coefficient (R2) in the range of 0.93–1 and RMSE was found to be near zero. The simulation results demonstrated that GP model is a good tool to predict the CI engine performance and emission parameters.
The valorization of waste materials is a viable alternative to traditional disposal systems, including in the field of renewable energy, biofuels and biomass ...
In recent years a large array of treatment protocols conceptualized as transdiagnostic have been developed with clinical and practical advantages compared to traditional single-disorder protocols. ...Within this panorama, the Transdiagnostic Unified Protocol (UP) of Emotional Disorders was developed aimed at treating the negative affective processes underlying several diagnostic categories, and accounting for the covariance of different emotional disorders. The UP has been found to efficiently target the roots of these disorders leading to a reduction in symptoms of co-occurring disorders. However, several questions have marginally addressed in the previous studies, and some UP features still remain unexplored. The present meta-analysis aims at evaluating whether the UP results to significant changes in anxiety and depression symptoms severity in children, adolescents, and adults. 19 RCTs and 13 uncontrolled pre-post trials comprising 2183 patients/clients met inclusion criteria for meta-analysis. Large to moderate combined overall effect size for both depression plus anxiety were detected in the uncontrolled pre-post studies (g = 0.756) and in RTCs studies (g = 0.452), respectively. Large effect size at pre-treatment to 3–6-month follow-up was observed for combined depression plus anxiety (g = 1.113). Subgroup analysis suggested that UP treatment does not differ across the anxiety and depression self-report measures. Moreover, UP intervention outperformed both passive and active control conditions to treat negative affective syndromes. Meta-regression confirmed the moderate effects of therapist level of experience, the sample characteristics, and the UP-protocol adaptations. The findings indicate that the manualized UP treatment has potential to contribute to improving mental health outcomes, particularly of anxiety and depression.
COVID-19 has rapidly emerged as a pandemic infection that has caused significant mortality and economic losses. Potential therapies and prophylaxis against COVID-19 are urgently needed to combat this ...novel infection. As a result of
evidence suggesting zinc sulphate may be efficacious against COVID-19, our hospitals began using zinc sulphate as add-on therapy to hydroxychloroquine and azithromycin.
To compare outcomes among hospitalized COVID-19 patients ordered to receive hydroxychloroquine and azithromycin plus zinc sulphate versus hydroxychloroquine and azithromycin alone.
This was a retrospective observational study. Data was collected from medical records for all patients with admission dates ranging from 2 March 2020 through to 11 April 2020. Initial clinical characteristics on presentation, medications given during the hospitalization, and hospital outcomes were recorded. The study included patients admitted to any of four acute care NYU Langone Health Hospitals in New York City. Patients included were admitted to the hospital with at least one positive COVID-19 test and had completed their hospitalization. Patients were excluded from the study if they were never admitted to the hospital or if there was an order for other investigational therapies for COVID-19.
Patients taking zinc sulphate in addition to hydroxychloroquine and azithromycin (
=411) and patients taking hydroxychloroquine and azithromycin alone (
=521) did not differ in age, race, sex, tobacco use or relevant comorbidities. The addition of zinc sulphate did not impact the length of hospitalization, duration of ventilation or intensive care unit (ICU) duration. In univariate analyses, zinc sulphate increased the frequency of patients being discharged home, and decreased the need for ventilation, admission to the ICU and mortality or transfer to hospice for patients who were never admitted to the ICU. After adjusting for the time at which zinc sulphate was added to our protocol, an increased frequency of being discharged home (OR 1.53, 95 % CI 1.12-2.09) and reduction in mortality or transfer to hospice among patients who did not require ICU level of care remained significant (OR 0.449, 95 % CI 0.271-0.744).
This study provides the first
evidence that zinc sulphate may play a role in therapeutic management for COVID-19.
Display omitted
•Urban microclimate tools are effective to test the impacts of climate adaptation strategies.•Out of 134 papers, 18 microclimate tools are deeply reviewed.•Their features and ...capabilities are compared and contrasted for a supporting tool selection.•Reviewed microclimate simulation tools show limitations and adopt strong assumptions.•Assumptions and limitations must be accounted for a reliable use of these tools.
The impact of human activities on climate change has become increasingly evident, with cities being particularly vulnerable to its effects. Anthropogenic emissions, such as heat and greenhouse gases, are projected to intensify climate-induced phenomena, which can lead to negative health outcomes. To understand how human health would be affected by such climate-exacerbated phenomena, computational models that consider the local microclimate are essential to better regulate cities to respond to these phenomena. Many simulation tools have been created and enhanced over the years. Therefore, this study systematically reviews the currently available urban microclimate simulation tools and compares their features and capabilities. The review suggests that these models can effectively assist in investigating urban health and testing adaptation strategies, but it is important to acknowledge their limitations due to assumptions made. Nonetheless, with proper interpretation and utilization, these models can provide valuable insights and contribute to informed decision-making processes.