In their paper, Kreiman et al. examined what it takes for an artificial neural network to be able to perform well on a new task without forgetting its previous knowledge. By comparing methods that stop task forgetting, they found that longer training times and maintenance of the most important connections in a particular task while training on a new one helped the neural network maintain its performance on both tasks. The authors hope that this proof-of-principle research will someday contribute to artificial intelligence that better mimics natural human intelligence.

The purpose of this study was to test devices installed on a gabled roof to see which reduced the actual uplift forces best. Three gabled birdhouse roofs were each modified with different mitigation devices: a rounded edge, a barrier shape, or an airfoil. The barrier edge had no significant effect on the time for the roof to blow off. The addition of airfoil devices on roofs, specifically in areas that are prone to hurricanes such as Florida, could keep roofs in place during hurricanes, thus reducing insurance bills, overall damage costs, and the loss of lives.

The use of fertilizers is associated with an increase in soil degradation, which is predicted to lead to a decrease in crop production within the next decade. Thus, it is critical to find solutions to support crop production to sustain the robust global population. In this study, the authors investigate how probiotic bacteria, like Rhizobium leguminosarum, Bacillus subtilis and Pseudomonas fluorescens, can impact the growth of Arabidopsis thaliana when applied to the seeds.They hypothesized that solutions with multiple bacterial species compared to those with only a single bacterial species would promote seed germination more effectively.

Acid rain has caused devastating decreases in ecosystems across the globe. To mimic the effect of acid rain on the environment, the authors analyzed the growth of gram-negative (Escherichia coli) and gram-positive (Staphylococcus epidermidis) bacteria in agar solutions with different pH levels. Results show that in a given acidic environment there was a significant decrease in bacterial growth with an increase in vinegar concentration in the agar, suggesting that bacterial growth is impacted by the pH of the environment. Therefore, increased levels of acid rain could potentially harm the ecosystem by altering bacterial growth.

The application of machine learning techniques has facilitated the automatic annotation of behavior in video sequences, offering a promising approach for ethological studies by reducing the manual effort required for annotating each video frame. Nevertheless, before solely relying on machine-generated annotations, it is essential to evaluate the accuracy of these annotations to ensure their reliability and applicability. While it is conventionally accepted that there cannot be a perfect annotation, the degree of error associated with machine-generated annotations should be commensurate with the error between different human annotators. We hypothesized that machine learning supervised with adequate human annotations would be able to accurately predict body parts from video sequences. Here, we conducted a comparative analysis of the quality of annotations generated by humans and machines for the body parts of sheep during treadmill walking. For human annotation, two annotators manually labeled six body parts of sheep in 300 frames. To generate machine annotations, we employed the state-of-the-art pose-estimating library, DeepLabCut, which was trained using the frames annotated by human annotators. As expected, the human annotations demonstrated high consistency between annotators. Notably, the machine learning algorithm also generated accurate predictions, with errors comparable to those between humans. We also observed that abnormal annotations with a high error could be revised by introducing Kalman Filtering, which interpolates the trajectory of body parts over the time series, enhancing robustness. Our results suggest that conventional transfer learning methods can generate behavior annotations as accurate as those made by humans, presenting great potential for further research.

The goal of this project was to see if the addition of wild orange essential oil to freshly squeezed orange juice would help to slow down the decay of ascorbic acid when exposed to various temperatures, allowing vital nutrients to be maintained and providing a natural alternative to the chemical additives in use in industry today. The authors hypothesized that the addition of wild orange essential oil to freshly squeezed orange juice would slow down the rate of oxidation when exposed to various temperatures, reducing ascorbic acid decay. On average, wild orange EO slowed down ascorbic acid decay in freshly squeezed orange juice by 15% at the three highest temperatures tested.

The Scripps National Spelling Bee (SNSB) is an iconic academic competition for United States (US) schoolchildren, held annually since 1925. However, the sizes and geographic distributions of sponsored regions are uneven. One state may send more than twice as many spellers as another state, despite similar numbers in child population. In 2018, the SNSB introduced a wildcard program known as RSVBee, which allowed students to apply to compete as a national finalist, even if they did not win their regional spelling bee. In this study, the authors tested the hypothesis that the geographic distribution of SNSB national finalists more closely matched the child population of the US after RSVBee was implemented.

Each day we are flooded with new items that promise us a better experience at a better price. This forces buyers to continuously chose between sticking to what they know, or trying something new. In turn, companies need to be aware of the factors affecting consumer choices, that too within the different fractions of society. In this study the authors investigate the effect of survey-based price setting on profits made based on African American teen purchases, and how African-American teen loyalty to a particular brand affects their willingness to pay a higher price than the market average for their preferred brand items.

Neural machine translation (NMT) is a software that uses neural network techniques to translate text from one language to another. However, one of the most famous NMT models—Google Translate—failed to give an accurate English translation of a famous Korean nursery rhyme, "Airplane" (비행기). The authors fine-tuned a pre-trained model first with a dataset from the lyrics domain, and then with a smaller dataset containing the rhythmical properties, to teach the model to translate rhythmically accurate lyrics. This stacked fine-tuning method resulted in an NMT model that could maintain the rhythmical characteristics of lyrics during translation while single fine-tuned models failed to do so.

Gun violence has been a serious issue in the South Side of Chicago for a long time. To intervene, regulators have passed legislation they hoped to curb -if not completely eradicate- the issue. However, there is little analysis done on how effective the various laws have been at reducing gun violence. Here the authors explore the association between firearm purchase laws passed between 1993-2018 and the incidence of gun homicide in Chicago's South Side. Their analysis suggests that some laws have been more effective than others, while some might have exacerbated the issue. However, they do not consider other contributing factors, which makes it difficult to prove causation without further investigation.