ALBERT

Description: Tail biting causes widespread problems both for animal welfare and in the form of economic losses in pig production. This study was performed to better understand the perceptions of farmers on how to best prevent tail biting, and if perceptions are influenced by the specific system of farming, with a focus on different levels of bedding use and docking different proportions of the tail of their pigs. Pig producers in the UK were surveyed on their perceptions of the efficacy of preventive measures and attitudes towards tail biting and docking. In total, 204 responses were included. The results show that producers rank the importance of preventive measures differently to scientists and other experts. This calls for consideration when communicating with producers; and for better integration of knowledge based on practical experiences with scientific results. The study also shows that the perception of how to best avoid tail biting differs between farms of different types, and that these perceptions might be influenced by the farmers´ own experiences—one example being that farms currently using plentiful amounts of bedding also value this more highly as a way to avoid tail biting than those that do not.

Description: Veterinary students face several ethical challenges during their curriculum. We used the Animal Ethics Dilemma to study animal ethical views of Finnish veterinary students, and also asked them to score the level of pain perception in 13 different species. Based on the 218 respondents, the utilitarian view was the dominating ethical view. Mammals were given higher pain scores than other animals. The proportion of the respect for nature view correlated negatively, and that of the animal rights view positively, with most animal pain scores. Fifth year students had a higher percentage of contractarian views, as compared to 1st and 3rd year students, but this might have been confounded by their age. Several pain perception scores increased with increasing study years. We conclude that the utilitarian view was clearly dominating, and that ethical views differed only slightly between students at different stages of their studies. Higher pain perception scores in students at a later stage of their studies might reflect an increased knowledge of animal capacities.

Description: Pecking-related problems are common in intensive egg production, compromising hen welfare, causing farmers economic losses and negatively affecting sustainability. These problems are often controlled by beak trimming, which in Finland is prohibited. An online questionnaire aimed to collect information from farmers about pecking-related problems in Finnish laying hen flocks, important risk factors and the best experiences to prevent the problems. Additionally, the farmers’ attitudes towards beak trimming were examined. We received 35 responses, which represents about 13% of all Finnish laying hen farms with ≥300 laying hens. The majority of respondents stated that a maximum of 5–7% incidence of feather pecking or 1–2% incidence of cannibalism would be tolerable. The majority of respondents (74%) expressed that they would definitely not use beak-trimmed hens. Only two respondents indicated that they would probably use beak-trimmed hens were the practice permitted. Among risk factors, light intensity earned the highest mean (6.3), on a scale from 1 (not important) to 7 (extremely important). Other important problems included those that occurred during rearing, feeding, flock management and problems with drinking water equipment (mean 5.9, each). The most important intervention measures included optimal lighting and feeding, flock management, and removing the pecker and victim. Concluding, Finnish farmers had strong negative attitudes towards beak trimming. The study underlines the importance of flock management, especially lighting and feeding, in preventing pecking problems and indicates that it is possible to incorporate a non-beak-trimming policy into sustainable egg production.

Description: High-quality data is necessary for modern machine learning. However, the acquisition of such data is difficult due to noisy and ambiguous annotations of humans. The aggregation of such annotations to determine the label of an image leads to a lower data quality. We propose a data-centric image classification benchmark with ten real-world datasets and multiple annotations per image to allow researchers to investigate and quantify the impact of such data quality issues. With the benchmark we can study the impact of annotation costs and (semi-)supervised methods on the data quality for image classification by applying a novel methodology to a range of different algorithms and diverse datasets. Our benchmark uses a two-phase approach via a data label improvement method in the first phase and a fixed evaluation model in the second phase. Thereby, we give a measure for the relation between the input labeling effort and the performance of (semi-)supervised algorithms to enable a deeper insight into how labels should be created for effective model training. Across thousands of experiments, we show that one annotation is not enough and that the inclusion of multiple annotations allows for a better approximation of the real underlying class distribution. We identify that hard labels can not capture the ambiguity of the data and this might lead to the common issue of overconfident models. Based on the presented datasets, benchmarked methods, and analysis, we create multiple research opportunities for the future directed at the improvement of label noise estimation approaches, data annotation schemes, realistic (semi-)supervised learning, or more reliable image collection.