Human Activity Recognition: A Dynamic Inductive Bias Selection Perspective

In this article, we study activity recognition in the context of sensor-rich environments. In these environments, many different constraints arise at various levels during the data generation process, such as the intrinsic characteristics of the sensing devices, their energy and computational constraints, and their collective (collaborative) dimension. These constraints have a fundamental impact on the final activity recognition models as the quality of the data, its availability, and its reliability, among other things, are not ensured during model deployment in real-world configurations. Current approaches for activity recognition rely on the activity recognition chain which defines several steps that the sensed data undergo: This is an inductive process that involves exploring a hypothesis space to find a theory able to explain the observations. For activity recognition to be effective and robust, this inductive process must consider the constraints at all levels and model them explicitly. Whether it is a bias related to sensor measurement, transmission protocol, sensor deployment topology, heterogeneity, dynamicity, or stochastic effects, it is essential to understand their substantial impact on the quality of the data and ultimately on activity recognition models. This study highlights the need to exhibit the different types of biases arising in real situations so that machine learning models, e.g., can adapt to the dynamicity of these environments, resist sensor failures, and follow the evolution of the sensors’ topology. We propose a metamodeling approach in which these biases are specified as hyperparameters that can control the structure of the activity recognition models. Via these hyperparameters, it becomes easier to optimize the inductive processes, reason about them, and incorporate additional knowledge. It also provides a principled strategy to adapt the models to the evolutions of the environment. We illustrate our approach on the SHL dataset, which features motion sensor data for a set of human activities collected in real conditions. The obtained results make a case for the proposed metamodeling approach; noticeably, the robustness gains achieved when the deployed models are confronted with the evolution of the initial sensing configurations. The trade-offs exhibited and the broader implications of the proposed approach are discussed with alternative techniques to encode and incorporate knowledge into activity recognition models.

The COVID-19 pandemic and the menstrual cycle: research gaps and opportunities

Since the beginning of the COVID-19 pandemic, discussions on social media and blogs have indicated that women have experienced menstrual changes, including altered menstrual duration, frequency, regularity, and volume (heavier bleeding and clotting), increased dysmenorrhea, and worsened premenstrual syndrome. There have been a small number of scientific studies of variable quality reporting on menstrual cycle features during the pandemic, but it is still unclear whether apparent changes are due to COVID-19 infection/illness itself, or other pandemic-related factors like increased psychological stress and changes in health behaviours. It is also unclear to what degree current findings are explained by reporting bias, recall bias, selection bias and confounding factors. Further research is urgently needed. We provide a list of outstanding research questions and potential approaches to address them. Findings can inform policies to mitigate against gender inequalities in health and society, allowing us to build back better post-COVID.

Métodos para Avaliação Global de Risco em Saúde Ocupacional: sabemos o suficiente?

Introduction / background / objectives Risk assessment in the workplace is a fundamental step towards obtaining safer and healthier jobs. The Security Technicians are generally the most experienced in this context; however, not all the professionals that carry out Occupational Health activities present well-structured and/ or practical knowledge about most of these methods. The purpose of this review was to summarize the main techniques used in this context. Methodology This is a review, initiated through a survey conducted in April 2020, in the RCAAP database (Open Access Scientific Repositories in Portugal). Content The author made some practical considerations about MARAT (Methodology for Risk Assessment and Accidents at Work), William Fine, MIAR (Integrated Methodology for Risk Assessment) and FMEA (Failure Mode and Effect Analysis), valuing with explanatory tables and highlighted the slight discrepancies between the documents consulted. Conclusions The key-words used were related to the methods that the author briefly know; in the documents found, sometimes, other techniques have been included; this obviously implies a bias selection. We easily find articles in indexed databases that mention these methods, but due to the limits imposed by most journals, relating to the size of the document, almost all authors only mention the name of the method or, at most, use a very synthetic description of it. In turn, in some Master’s or Doctorate Theses (where this problem does not exist) we can find a more methodological description, but still, sometimes you cannot always get the practice knowledge of how to use all methods or if the items are slightly different, result of adaptations, consideration of different subtypes or a mixture of methods, carried out over the decades. Any professional on an Occupational Health Team will have a reasonable sense of what the most damaging tasks will be; however, presenting this evidence, attenuating subjectivity and making use of the hierarchy that mathematical scales can offer, it becomes more accepted as valid by employers/ representatives/ workers and, consequently, increase the receptivity to proposed measures to mitigate/ correct the problem and reassess it, after introducing corrective measures. It would be desirable for all professionals in the field to have (at least) a generic idea of ​​the existing methods and where they can seek more information, in order to execute these techniques, when necessary.

Shooting for neutrality? Analysing bias in terrorism reports in Dutch newspapers

Similar to other nations, terrorism is a compelling preoccupation in the Netherlands. One issue in the public debate concerning news coverage is whether it fairly reports the perpetrators’ racial, ethnic and religious backgrounds. This article asks whether there is disproportionate attention (coverage bias), selection (gatekeeping bias) and presentation (statement bias) in various Dutch newspapers between 2015 and 2017. Using content analysis, the authors find all three types of bias present, albeit to different degrees. We propose that Critical Race Theory (CRT) usefully explains how bias is often unintentional and that journalistic outcomes are the consequence of unconsciously imprinted ideas about what constitutes a ‘terrorist’, facilitated and amplified by institutionalized media practices and wider societal power relations.

Micro-Loans and Household Economies: Evidence in Indonesia

Micro-loans intended to improve household economies are a fascinating subject for research because a comparative analysis of before and after taking micro-loans would result in a bias selection. Households have different prior conditions from one another, so the difference found during the study is not entirely due to receiving micro-loans. There is a risk of moral hazard risk due to asymmetric information. This research adopts the double difference (DD) fixed effects method to estimate the extent of micro-loans’ impact. Results indicate that micro-loans are significantly influencing the household economies. The impact size was relatively small that it was not apparent during regression. As an implication, micro-loans intended for productive purposes can help improve household economic conditions. Effective and sustainable monitoring and counsel can minimize the risk of moral hazard.DOI: 10.15408/sjie.v7i1.5954

The male handicap: male-biased mortality explains skewed sex ratios in brown trout embryos

Juvenile sex ratios are often assumed to be equal for many species with genetic sex determination, but this has rarely been tested in fish embryos due to their small size and absence of sex-specific markers. We artificially crossed three populations of brown trout and used a recently developed genetic marker for sexing the offspring of both pure and hybrid crosses. Sex ratios (SR = proportion of males) varied widely one month after hatching ranging from 0.15 to 0.90 (mean = 0.39 ± 0.03). Families with high survival tended to produce balanced or male-biased sex ratios, but SR was significantly female-biased when survival was low, suggesting that males sustain higher mortality during development. No difference in SR was found between pure and hybrid families, but the existence of sire × dam interactions suggests that genetic incompatibility may play a role in determining sex ratios. Our findings have implications for animal breeding and conservation because skewed sex ratios will tend to reduce effective population size and bias selection estimates.