RACOG and wRACOG: Two Probabilistic Oversampling Techniques.

As machine learning techniques mature and are used to tackle complex scientific problems, challenges arise such as the imbalanced class distribution problem, where one of the target class labels is under-represented in comparison with other classes. Existing oversampling approaches for addressing this problem typically do not consider the probability distribution of the minority class while synthetically generating new samples. As a result, the minority class is not well represented which leads to high misclassification error. We introduce two Gibbs sampling-based oversampling approaches, namely RACOG and wRACOG, to synthetically generating and strategically selecting new minority class samples. The Gibbs sampler uses the joint probability distribution of attributes of the data to generate new minority class samples in the form of Markov chain. While RACOG selects samples from the Markov chain based on a predefined lag, wRACOG selects those samples that have the highest probability of being misclassified by the existing learning model. We validate our approach using five UCI datasets that were carefully modified to exhibit class imbalance and one new application domain dataset with inherent extreme class imbalance. In addition, we compare the classification performance of the proposed methods with three other existing resampling techniques.

Activity Recognition on Streaming Sensor Data.

Many real-world applications that focus on addressing needs of a human, require information about the activities being performed by the human in real-time. While advances in pervasive computing have lead to the development of wireless and non-intrusive sensors that can capture the necessary activity information, current activity recognition approaches have so far experimented on either a scripted or pre-segmented sequence of sensor events related to activities. In this paper we propose and evaluate a sliding window based approach to perform activity recognition in an on line or streaming fashion; recognizing activities as and when new sensor events are recorded. To account for the fact that different activities can be best characterized by different window lengths of sensor events, we incorporate the time decay and mutual information based weighting of sensor events within a window. Additional contextual information in the form of the previous activity and the activity of the previous window is also appended to the feature describing a sensor window. The experiments conducted to evaluate these techniques on real-world smart home datasets suggests that combining mutual information based weighting of sensor events and adding past contextual information into the feature leads to best performance for streaming activity recognition.

Transfer Learning for Activity Recognition: A Survey.

Many intelligent systems that focus on the needs of a human require information about the activities being performed by the human. At the core of this capability is activity recognition, which is a challenging and well-researched problem. Activity recognition algorithms require substantial amounts of labeled training data yet need to perform well under very diverse circumstances. As a result, researchers have been designing methods to identify and utilize subtle connections between activity recognition datasets, or to perform transfer-based activity recognition. In this paper we survey the literature to highlight recent advances in transfer learning for activity recognition. We characterize existing approaches to transfer-based activity recognition by sensor modality, by differences between source and target environments, by data availability, and by type of information that is transferred. Finally, we present some grand challenges for the community to consider as this field is further developed.

Understanding calibration of deep neural networks for medical image classification.

Background and Objective - In the field of medical image analysis, achieving high accuracy is not enough; ensuring well-calibrated predictions is also crucial. Confidence scores of a deep neural network play a pivotal role in explainability by providing insights into the model's certainty, identifying cases that require attention, and establishing trust in its predictions. Consequently, the significance of a well-calibrated model becomes paramount in the medical imaging domain, where accurate and reliable predictions are of utmost importance. While there has been a significant effort towards training modern deep neural networks to achieve high accuracy on medical imaging tasks, model calibration and factors that affect it remain under-explored. Methods - To address this, we conducted a comprehensive empirical study that explores model performance and calibration under different training regimes. We considered fully supervised training, which is the prevailing approach in the community, as well as rotation-based self-supervised method with and without transfer learning, across various datasets and architecture sizes. Multiple calibration metrics were employed to gain a holistic understanding of model calibration. Results - Our study reveals that factors such as weight distributions and the similarity of learned representations correlate with the calibration trends observed in the models. Notably, models trained using rotation-based self-supervised pretrained regime exhibit significantly better calibration while achieving comparable or even superior performance compared to fully supervised models across different medical imaging datasets. Conclusion - These findings shed light on the importance of model calibration in medical image analysis and highlight the benefits of incorporating self-supervised learning approach to improve both performance and calibration.

Digital screening and brief intervention for illicit drug misuse in college students: A mixed methods, pilot, cluster, randomized trial from India.

BACKGROUND: Adolescence and early adulthood are vulnerable periods for substance use-related disorders later in life. The use of internet-enabled interventions can be useful, especially in low-resource settings. AIMS: To examine the feasibility, acceptability, and preliminary effectiveness of single-session digital screening and brief intervention (d-SBI) for illicit drug misuse in college students and explore barriers and facilitators of d-SBI. METHODS: Design: Mixed-methods, pilot cluster randomized trial. SETTING: Four conveniently selected colleges were randomized into intervention and control groups. PARTICIPANTS: 219 students were screened, and 37 fulfilled eligibility. Twenty-four completed follow-ups. In-depth interviews were done with ten students. Intervention and Comparator: Following a digital screening, Alcohol, Smoking, and Substance Involvement Screening Test (ASSIST) based brief intervention was provided in the d-SBI group. The control group received brief education. MEASUREMENTS: Acceptability was assessed by direct questions and usage statistics. ASSIST scores of groups were assessed at baseline and 3 months. Inductive coding of the interview transcript was done. RESULTS: More than 50 % of participants found d-SBI user-friendly, appropriate, and useful. Eighty percent of users, who logged in, completed screening. Per-protocol analysis showed a reduction in cannabis-ASSIST score over 3 months. The mean ASSIST score for other drugs combined did not differ significantly between groups. The difference in risk transition (moderate to low) was not significant. Qualitative analysis revealed three overarching themes- recruitment, engagement, and behavior change. CONCLUSIONS: Digital SBI for drug misuse is feasible among college students. d-SBI might be effective in reducing cannabis use.

Digital screening and brief intervention for alcohol misuse in college students: A pilot, mixed-methods, cluster randomized controlled trial from a low-resourced setting.

INTRODUCTION: We examined the feasibility and acceptability of digital screening and brief intervention (d-SBI) for alcohol misuse in college students; the effectiveness of d-SBI was our secondary outcome. We also explored the barriers and facilitators of d-SBI. METHODS: The study design is a mixed-methods, pilot, and cluster randomized trial. Five colleges from a northern city in India were randomly allocated to d-SBI and control groups. One hundred and ninety-one students were screened, and 25 (male = 23 and female = 2) participants (age 19.62 ± 2.58 years) fulfilled eligibility. All participants completed follow-up assessments at 3 months. In-depth interviews were done with 11 participants. Alcohol Use Disorder Identification Test (AUDIT) based screening brief intervention was provided on a web portal- or mobile application in the d-SBI group. The control group received digital screening and brief education. Direct questions and usage statistics assessed the measurement acceptability of the intervention. We compared the change in AUDIT scores in the intervention groups over 3 months post-intervention. Thematic analyses of transcripts of interviews were done by inductive coding. RESULTS: Most participants reported that d-SBI was user-friendly (80%), advice was appropriate (80%), and perceived it to be useful (72%). Ninety-six percent of users, who logged in, completed screening. There was a significant decrease in AUDIT scores both in d-SBI (p < .001) and control groups (p < .001). Time and group significantly affected the mean AUDIT score, but time × group interaction was non-significant. Thematic analysis revealed six overarching themes. CONCLUSIONS: Digital SBI for alcohol misuse is acceptable, feasible, and possibly effective among college students from low-resource settings.

Machine learning methods trained on simple models can predict critical transitions in complex natural systems.

Forecasting sudden changes in complex systems is a critical but challenging task, with previously developed methods varying widely in their reliability. Here we develop a novel detection method, using simple theoretical models to train a deep neural network to detect critical transitions-the Early Warning Signal Network (EWSNet). We then demonstrate that this network, trained on simulated data, can reliably predict observed real-world transitions in systems ranging from rapid climatic change to the collapse of ecological populations. Importantly, our model appears to capture latent properties in time series missed by previous warning signals approaches, allowing us to not only detect if a transition is approaching, but critically whether the collapse will be catastrophic or non-catastrophic. These novel properties mean EWSNet has the potential to serve as an indicator of transitions across a broad spectrum of complex systems, without requiring information on the structure of the system being monitored. Our work highlights the practicality of deep learning for addressing further questions pertaining to ecosystem collapse and has much broader management implications.

Can google trends search inform us about the population response and public health impact of abrupt change in alcohol policy? A case study from India during the covid-19 pandemic.

BACKGROUND: Sudden alcohol prohibition in India during the COVID-19 pandemic presented an opportunity to test whether Google Trends data could indicate population responses and the public health impact of alcohol policy. We hypothesized, following prohibition: there would be a significant change in the relative search volumes (RSV) of alcohol-related queries; that temporal analysis of the trends would reflect a public response to policy changes; and that geospatial analysis of RSV would correlate with the prevalence of alcohol use. METHODS: Three different search periods were used to test the hypotheses. The search inputs were based on potential public response to alcohol prohibition, as evidenced by the literature, newspaper articles, and consensus. We used RSV as the unit of analysis. Mean RSV of search queries, pre-post implementation of prohibition, were compared. Smoothing of scatter plots examined the temporal association of trends with policy measures. Multiple linear regression tested the relationship of state-wise RSV and alcohol use prevalence. RESULTS: Post-implementation of prohibition, a significant increase in the RSV was observed for searches related to alcohol withdrawal (p<0.001), how to extract alcohol from sanitizer (p = 0.002), alcohol home delivery online (p<0.001), alcohol home delivery (p<0.001), and sleeping pills (p = 0.006). The trends suggested a decrease in general interest in alcohol but increased demand, and a possible connection with changes in policy measures. State-level RSV and alcohol use prevalence did not reveal a significant relationship. CONCLUSION: Google trend is a potential source of rapid feedback to policymakers about population responses to an abrupt change in alcohol policies.

One-Class Classification-Based Real-Time Activity Error Detection in Smart Homes.

Caring for individuals with dementia is frequently associated with extreme physical and emotional stress, which often leads to depression. Smart home technology and advances in machine learning techniques can provide innovative solutions to reduce caregiver burden. One key service that caregivers provide is prompting individuals with memory limitations to initiate and complete daily activities. We hypothesize that sensor technologies combined with machine learning techniques can automate the process of providing reminder-based interventions. The first step towards automated interventions is to detect when an individual faces difficulty with activities. We propose machine learning approaches based on one-class classification that learn normal activity patterns. When we apply these classifiers to activity patterns that were not seen before, the classifiers are able to detect activity errors, which represent potential prompt situations. We validate our approaches on smart home sensor data obtained from older adult participants, some of whom faced difficulties performing routine activities and thus committed errors.

Activity discovery and activity recognition: a new partnership.

Activity recognition has received increasing attention from the machine learning community. Of particular interest is the ability to recognize activities in real time from streaming data, but this presents a number of challenges not faced by traditional offline approaches. Among these challenges is handling the large amount of data that does not belong to a predefined class. In this paper, we describe a method by which activity discovery can be used to identify behavioral patterns in observational data. Discovering patterns in the data that does not belong to a predefined class aids in understanding this data and segmenting it into learnable classes. We demonstrate that activity discovery not only sheds light on behavioral patterns, but it can also boost the performance of recognition algorithms. We introduce this partnership between activity discovery and online activity recognition in the context of the CASAS smart home project and validate our approach using CASAS data sets.

CASAS: A Smart Home in a Box.

While the potential benefits of smart home technology are widely recognized, a lightweight design is needed for the benefits to be realized at a large scale. We introduce the CASAS "smart home in a box", a lightweight smart home design that is easy to install and provides smart home capabilities out of the box with no customization or training. We discuss types of data analysis that have been performed by the CASAS group and can be pursued in the future by using this approach to designing and implementing smart home technologies.

Hierarchical domain adaptation for SEMG signal classification across multiple subjects.

Large variations in Surface Electromyogram (SEMG) signal across different subjects make the process of automated signal classification as a generalized tool, challenging. In this paper, we propose a domain adaptation methodology that addresses this challenge. In particular we propose a hierarchical sample selection methodology, that selects samples from multiple training subjects, based on their similarity with the target subject at different levels of granularity. We have validated our framework on SEMG data collected from 8 people during a fatiguing exercise. Comprehensive experiments conducted in the paper demonstrate that the proposed method improves the subject independent classification accuracy by 21% to 23% over the cases without domain adaptation methods and by 14% to 20% over the existing state-of-the-art domain adaptation methods.