Search for: All records

Machine Learning (ML) is increasingly gaining significance for end-user programmer (EUP) applications. However, machine learning end-user programmers (ML-EUPs) without the right background face a daunting learning curve and a heightened risk of mistakes and flaws in their models. In this work, we designed a conversational agent named “Newton” as an expert to support ML-EUPs. Newton’s design was shaped by a comprehensive review of existing literature, from which we identified six primary challenges faced by ML-EUPs and five strategies to assist them. To evaluate the efficacy of Newton’s design, we conducted a Wizard of Oz within-subjects study with 12 ML-EUPs. Our findings indicate that Newton effectively assisted ML-EUPs, addressing the challenges highlighted in the literature. We also proposed six design guidelines for future conversational agents, which can help other EUP applications and software engineering activities. 

Semiconductor InSe 2D nanomaterials have emerged as potential photoresponsive materials for broadly distributed photodetectors and wearable electronics technologies due to their high photoresponsivity and thermal stability. This paper addresses an environmental concern about the fate of InSe 2D nanosheets when disposed and released into the environment after use. Semiconducting materials are potentially reactive and often form environmentally damaging species, for example reactive oxygen and nitrogen species, when degraded. InSe nanosheets are prepared using a semi bottom-up approach which involves a reaction between indium and selenium precursors at elevated temperature in an oxygen-free environment to prevent oxidation. InSe nanosheets are formed as a stable intermediate with micrometer-sized lateral dimensions and a few monolayer thickness. The InSe 2D nanosheets are obtained when the reaction is stopped after 30 minutes by cooling. Keeping the reaction at elevated temperature for a longer period, for example 60 minutes leads to the formation of InSe 3D nanoparticles of about 5 nm in diameter, a thermodynamically more stable form of InSe. The paper focuses on the colloidal stabilization of InSe nanosheets in an aqueous solution that contains epigallocatechin gallate (EGCG), a natural organic matter (NOM) simulant. We show that EGCG coats the surface of the hydrophobic, water-insoluble InSe nanosheets via physisorption. The formed EGCG-coated InSe nanosheets are colloidally stable in aqueous solution. While unmodified semiconducting InSe nanosheets could produce reactive oxygen species (ROS) when illuminated, our study shows low levels of ROS generation by EGCG-coated InSe nanosheets under ambient light, which might be attributed to ROS quenching by EGCG. Growth-based viability (GBV) assays show that the colloidally stable EGCG-coated InSe nanosheets adversely impact the bacterial growth of Shewanella oneidensis MR-1, an environmentally relevant Gram-negative bacterium in aqueous media. The impact on bacterial growth is driven by the EGCG coating of the nanosheets. In addition, live/dead assays show insignificant membrane damage of the Shewanella oneidensis MR-1 cells by InSe nanosheets, suggesting a weak association of EGCG-coated nanosheets with the cells. It is likely that the adverse impact of EGCG-coated nanosheets on bacterial growth is the result of increasing local concentration of EGCG either when adsorbed on the nanosheets when the nanosheets interact with the cells, or when desorbed from the EGCG-coated nanosheets to interact with the bacterial cells. 

Increasing temperatures have raised concerns over the potential effect on disease spread. Temperature is a well known factor affecting mosquito population dynamics and the development rate of the malaria parasite within the mosquito, and consequently, malaria transmission. A sinusoidal wave is commonly used to incorporate temperature effects in malaria models, however, we introduce a seasonal malaria framework that links data on temperature-dependent mosquito and parasite demographic traits to average monthly regional temperature data, without forcing a sinusoidal fit to the data. We introduce a spline methodology that maps temperature-dependent mosquito traits to time-varying model parameters. The resulting non-autonomous system of differential equations is used to study the impact of seasonality on malaria transmission dynamics and burden in a high and low malaria transmission region in Malawi. We present numerical simulations illustrating how temperature shifts alter the entomological inoculation rate and the number of malaria infections in these regions. 

Use of complex metal oxide nanoparticles has drastically risen in recent years, especially due to their utility in electric vehicle batteries. However, use of these materials has outpaced our understanding of how they might affect environmental organisms, which they could encounter through release during manufacture, use, and disposal. In particular, little is known about the effects of chronic exposure to complex metal oxide nanoparticles. Here, we have focused on an environmentally-relevant bacterial species, Shewanella oneidensis, which is ubiquitous in nature and responsible for bioremediation of heavy metals, and assessed the toxic effects of nanoscale lithiated nickel manganese cobalt oxide (NMC), which is an emerging battery cathode material for electronic devices. We previously reported that chronic exposure of S. oneidensis to NMC results in the emergence of an adaptive phenotype where the bacteria are able to tolerate otherwise lethal concentrations of NMC. In the present study, we aim to investigate the role of reactive oxygen species (ROS) and changes in phenotype of the NMC-adapted bacterial population. We found that NMC-exposed bacteria possess ROS-containing membrane vesicles, as well as an increased propensity to generate random DNA mutations and harbor other DNA damage. Thus, our data indicate substantial genetic-level variation in bacteria that results from chronic exposure to toxic complex metal oxide nanomaterials. 

Humans understand videos from both the visual and audio aspects of the data. In this work, we present a self supervised cross modal representation approach for learning audio visual correspondence (AVC) for videos in the wild. After the learning stage, we explore retrieval in both cross modal and intra modal manner with the learned representations. We verify our experimental results on the VGGSound dataset and our approach achieves promising results.