Search Results (7)

Federated learning (FL) enables collaborative model building among a large number of participants without sharing sensitive data to the central server. Because of its distributed nature, FL has limited control over local data and the corresponding training process. Therefore, it is susceptible to data poisoning attacks where malicious workers use malicious training data to train the model. Furthermore, attackers on the worker side can easily manipulate local data by swapping the labels of training instances, adding noise to training instances, and adding out-of-distribution training instances in the local data to initiate data poisoning attacks. And local workers under such attacks carry incorrect information to the server, poison the global model, and cause misclassifications. So, the prevention and detection of such data poisoning attacks is crucial to build a robust federated training framework. To address this, we propose a prevention strategy in federated learning, namely confident federated learning, to protect workers from such data poisoning attacks. Our proposed prevention strategy at first validates the label quality of local training samples by characterizing and identifying label errors in the local training data, and then excludes the detected mislabeled samples from the local training. To this aim, we experiment with our proposed approach on both the image and audio domains, and our experimental results validated the robustness of our proposed confident federated learning in preventing the data poisoning attacks. Our proposed method can successfully detect the mislabeled training samples with above 85% accuracy and exclude those detected samples from the training set to prevent data poisoning attacks on the local workers. However, our prevention strategy can successfully prevent the attack locally in the presence of a certain percentage of poisonous samples. Beyond that percentage, the prevention strategy may not be effective in preventing attacks. In such cases, detection of the attacked workers is needed. So, in addition to the prevention strategy, we propose a novel detection strategy in the federated learning framework to detect the malicious workers under attack. We propose to create a class-wise cluster representation for every participating worker by utilizing the neuron activation maps of local models and analyze the resulting clusters to filter out the workers under attack before model aggregation. We experimentally demonstrated the efficacy of our proposed detection strategy in detecting workers affected by data poisoning attacks, along with the attack types, e.g., label-flipping or dirty labeling. In addition, our experimental results suggest that the global model could not converge even after a large number of training rounds in the presence of malicious workers, whereas after detecting the malicious workers with our proposed detection method and discarding them from model aggregation, we ensured that the global model achieved convergence within very few training rounds. Furthermore, our proposed approach stays robust under different data distributions and model sizes and does not require prior knowledge about the number of attackers in the system. Full article

Within the dynamic, complex, and often safety-critical operations of many process industries, the integration of technology and human elements has given rise to sociotechnical systems (STSs), where the interaction between people and technology plays a pivotal role. To thrive in this complex environment, organizations must adopt adaptive error management strategies and cultivate organizational resilience. This approach involves managing the unexpected and designing systems to embrace disorder by organizational learning from errors in STSs. The main objective of this article was to present empirical data of error-causing elements in STSs based on the Dirty Dozen concept, their underlying structure, and implications for error causation screening and adaptive error management systems. A sample of 544 workers employed in seven process industries, such as automotive, chemicals, defense, metal, and timber, participated in this study. The results revealed a three-factor model of human error causation in STSs. Based on these results, an adaptive error management system (AEMS), which includes evidence-based interventions to manage causes of human errors and mitigate their risky consequences, was presented. Finally, implications for organizational resilience and safety culture in STSs were discussed. Full article

The literature on dirty workers analysed material and symbolic aspects of work, highlighting how dirty workers deal with the stigma associated with their occupations. This approach put less attention on dirty workers who operate in the informal economy, which is a relevant category especially in some sectors, such as the urban-waste management and recycling. Drawing on a 2019–2022 qualitative study of waste pickers (WPs) in the Rio Grande do Sul state (Brazil), this article aims to understand whether and how the informal conditions interact with the symbolic and material aspects of the dirty job. First, it shows various attempts to redefine the meanings of waste pickers’ work, in a positive sense. The more articulated attempts mainly concern the more structured WPs organisations and, in some cases, go far beyond the strategies described by the traditional literature on dirty workers. Second, it emphasizes the importance of the relations between WPs organisations and public administrations in defining the effectiveness of the actions aimed at reducing the stigma associated with the WPs’ work. Together, these contributions highlight the utility of exploring the dynamics and the differences of informal dirty work in order to enrich the dirty-work approach. Full article

Since the Fukushima Daiichi Nuclear Power Plant accident in March 2011, the technology available for unmanned aerial vehicles (UAVs) for radiation monitoring has improved greatly. Remote access to radiation-contaminated areas not only eliminates unnecessary exposure of civilians or military personnel, but also allows workers to explore inaccessible places. Hazardous levels of radioactive contamination can be expected as a result of accidents in the nuclear power industry or as a result of the intentional release of radioactive materials for terrorist purposes (dirty bombs, building contamination, etc.). The possibility to detect, identify, and characterize radiation and nuclear material using mobile and remote sensing platforms is a common requirement in the radiation sensing community. The technology has applications in homeland security and law enforcement, customs and border protection, nuclear power plant safety and security, nuclear waste monitoring, environmental recovery, and the military. In this work, the authors have developed, implemented, and characterized a gamma-ray detection and spectroscopy system capable of operating on a UAV. The system was mainly developed using open-source software and affordable hardware components to reduce development and maintenance costs and provide satisfactory performance as a detection instrument. The designed platform can be used to perform mapping or localization tasks to improve the risk assessment process for first responders during the management of radiological and nuclear incidents. First, the design process of the system is described; the result of the characterization of the platform is then presented together with the use of the prototype installed on a UAV in an exercise simulating a radiological and nuclear contamination scenario. Full article

Pre-COVID-19 pandemic, restaurant workers comprised one of the largest workforces in the United States, contributing hundreds of billions of dollars to the national economy. Yet, restaurant workers routinely face customer abuse, meager wages, lack of benefits, sexual harassment, and one of the highest rates of turnover across industries. Given these conditions, this qualitative study investigates how restaurant workers make sense of a contested occupation and manage the stigma associated with their occupation. Specifically, this study examines how food and beverage service workers identify with and navigate a demanding industry while managing the sociocultural assumptions of service work. Using a multi-level discourse analytic framework, we analyze how service workers craft and enact occupational identities. Through an analysis of in-depth interviews with 19 restaurant employees, we demonstrate how people foreground the positive attributes of restaurant work while resisting social Discourses that position the work as dirty, demeaning, emotional, and meaningless. We analyze how workers frame the values of working in restaurants and the communicative strategies they use to navigate stigmatized social interactions, including emphasizing flexibility, empathy, emotion management, and teamwork. Theoretical and practical implications offer suggestions to improve workforce sustainability and working conditions for employees. Full article

Migrants are mainly employed in “3D jobs” which are dirty, dangerous, and difficult, are characterized by monotony and intense rhythms, and are found in sectors with higher risks such as construction, heavy industry, and agriculture. The aim of this study is to construct a systematic review in order to identify the main occupational risks and occupational diseases of this category. Research included articles published from 2013 to 2018 on major online databases (PubMed, Cochrane Library, and Scopus), which were obtained using a combination of keywords (migrant workers, expatriates, physical health, diseases, illnesses, travel, travelers, work, and occupational medicine). The online search included 1109 references. We excluded 977 studies because they were unrelated to physical health and another 64 due to duplication. There were 68 articles which were analyzed, including six reviews and 62 original articles. The main risks which emerged are developing infectious diseases, metabolic cardiovascular diseases and manifesting a lower quality of life, in particular due to difficulties in accessing local health services. It will be crucial to implement the role of occupational medicine in order to introduce multilevel interventions designed to prevent work-related injuries and illnesses, and to promote healthier working environments. Full article

‘Electrolyzed water’ generators are readily available in the food industry as a renewable source of hypochlorous acid that eliminates the need for workers to handle hazardous hypochlorite concentrates. We applied electrolyzed water (EW) directly to multi-strain cocktails of Listeria monocytogenes, E. coli O157:H7, and Salmonella sp. at 250 ppm free available chlorine (FAC) and achieved greater than 6-log reductions in 2 min. Lower EW values were examined as antimicrobial interventions for fresh meat (beef carcasses), processed meats (frankfurters), and food contact surfaces (slicing blades). Little or no reduction relative to controls was observed when generic E. coli-inoculated beef carcasses or L. monocytogenes-inoculated frankfurters were showered with EW. Spray application of EW (25 and 250-ppm FAC) onto L. monocytogenes-inoculated slicing blades showed that greater reductions were obtained with ‘clean’ (3.6 and 5.7-log reduction) vs. ‘dirty’ (0.6 and 3.3-log reduction) slicing blades, respectively. Trials with L. monocytogenes-inoculated protein-EW solutions demonstrated that protein content as low as 0.1% is capable of eliminating FAC, reducing antimicrobial activity against L. monocytogenes. EW appears better positioned as a surface sanitizer with minimal organic material that can otherwise act as an effective reducing agent to the oxidizing solution rendering it ineffective. Full article