Do you want to publish a course? Click here

On the Impact of Random Seeds on the Fairness of Clinical Classifiers

على تأثير بذور عشوائية على نزاهة الطبقات السريرية

183   0   0   0.0 ( 0 )
 Publication date 2021
and research's language is English
 Created by Shamra Editor




Ask ChatGPT about the research

Recent work has shown that fine-tuning large networks is surprisingly sensitive to changes in random seed(s). We explore the implications of this phenomenon for model fairness across demographic groups in clinical prediction tasks over electronic health records (EHR) in MIMIC-III ------ the standard dataset in clinical NLP research. Apparent subgroup performance varies substantially for seeds that yield similar overall performance, although there is no evidence of a trade-off between overall and subgroup performance. However, we also find that the small sample sizes inherent to looking at intersections of minority groups and somewhat rare conditions limit our ability to accurately estimate disparities. Further, we find that jointly optimizing for high overall performance and low disparities does not yield statistically significant improvements. Our results suggest that fairness work using MIMIC-III should carefully account for variations in apparent differences that may arise from stochasticity and small sample sizes.



References used
https://aclanthology.org/
rate research

Read More

This study was carried out during 2014-2015 in two different nurseries located in University nursery (Lattakia governorate) and Salhab nursery (countryside of Hama governorate). Seeds were taken from two sites of Lattakia governorate (Mashqita– Bsn ada) where Cordia myxa exists as exotic species in some coastal regions and three treatments in addition to control were examined .The results showed that there is an effect of scarification of seeds on investigated treatments and were significant differences in germination percentage in both sites which recorded 55% in university nursery and 35% in Al-Ghab nursery. While there were not contrast between the soaking in boiling and tap water treatments on the seed germination energy in Al-Ghab nursery. Scarification treatment gave higher value of germination energy 25% in Al-Ghab nursery and 21.66% in university nursery , while the same treatment decreased germination velocity compared to other treatments in university nursery which seed needed in average ( 8,01 day) whereas the same treatment increased germination velocity in Al-Ghab nursery which needed (1,54 day). Finally, regarding to cutting ,There is not observe to any positive influence to cuttings which treated in ABA hormone (1g/l) for (10,20,30 m). It could be deduced that there were changes in germination percentage, germination energy and germination velocity in two sites which appeared to be a Preliminary indication of impact of the some dormancy breaking treatments on investigated indications of Cordia myxa Forsk. species .
The effect of different treatments on carob seed germination was studied. The results showed that, seed scarification was the most effective method in increasing germination rate as well as the use of concentrated sulfuric acid for 30 minutes reac hing 100%, followed by the treatment with alcohol for 60 minutes, and immersing seeds in hot water for 60 minutes. The results also indicated that the speed of germination when seeds were scarified was the best (I.D.=14.5), followed by the treatment with concentrated sulfuric acid for 30 minutes (I.D.=21.5), alcohol for 60 minutes (I.D.= 38) and immersing seeds in hot water for 60 minutes (ID.=60).
Low birth weight and premature infant need complementary treatment intervention to promote optimal clinical outcome and to decrease the immediate adversities and developmental deficits associated with prematurity and intensive care unit environment. The purpose of this study was to investigate the effect of complementary treatment program on the clinical outcome for preterm and low birth weight infant. The study was conducted at the pediatric intensive care units of AL-Assad hospital and maternal &child hospital (lattakia). The sample was divided into two groups of 20 infants experimental croup and 10 infant control croup with gestational age less than 36 weeks at birth, birth weight less than 2500g and more than 1500 g, and no congenital anomalies. The clinical outcome was significantly improved in clinical outcome, while no change in the control group. The experimental group had significantly weight gain, higher scores for awake state and motor activity than the control group. Significantly greater fidgeting or crying, and increased motor activity. The results of this study advice to apply complementary treatment therapy because its affect in improve clinical outcome. Nursing staff can use complementary treatment to promote the infant's capability to respond positively to his environment and to provide developmental support for healthy premature infants
This study aimed to identify the impact of budget participation on each of managerial performance, organizational commitment and distributive fairness, to investigate the impact of organizational commitment and distributive fairness on managerial p erformance, and to study the impact of distributive fairness on organizational commitment. To achieve the study objectives, a questionnaire was used and distributed on the Agricultural Firms in the Syrian Coast. This questionnaire was distributed to72 respondents, but only 50 questionnaires were completed and received. The results showed that budgetary participation has a positive impact on each of the manage rial performance, organizational commitment and distributive fairness. And a positive impact each of the Organizational Commitment and the Distributive Fairnesson the managerial performance. At last the results appeared that there is no significant impact for Distributive Fairnesson budget goal commitment.
The Shared Task on Hateful Memes is a challenge that aims at the detection of hateful content in memes by inviting the implementation of systems that understand memes, potentially by combining image and textual information. The challenge consists of three detection tasks: hate, protected category and attack type. The first is a binary classification task, while the other two are multi-label classification tasks. Our participation included a text-based BERT baseline (TxtBERT), the same but adding information from the image (ImgBERT), and neural retrieval approaches. We also experimented with retrieval augmented classification models. We found that an ensemble of TxtBERT and ImgBERT achieves the best performance in terms of ROC AUC score in two out of the three tasks on our development set.

suggested questions

comments
Fetching comments Fetching comments
Sign in to be able to follow your search criteria
mircosoft-partner

هل ترغب بارسال اشعارات عن اخر التحديثات في شمرا-اكاديميا