Illegal insider trading undermines the integrity of financial markets by exploiting non-public information for personal gain, posing risks to market fairness and investor trust. Addressing this issue is crucial to ensure transparency and maintain investor confidence. Estimates indicate that illegal insider trading occurs in about 20% of merger and acquisition events and 5% of quarterly earnings announcements. Our research introduces a deep learning-based approach that combines text classification, time series forecasting, anomaly detection and explainable AI to detect and predict potential illegal insider trading. By analyzing litigation related press releases and incorporating market and social media data, we develop a method to identify suspicious trading patterns. Our findings demonstrate that the proposed approach is promising in detecting potential instances of illegal insider trading, uncovering subtle patterns and anomalies. These insights offer financial regulators and institutions an enhanced surveillance mechanism, contributing to ongoing efforts to safeguard market fairness and reliability.

As malware threats continue to evolve in complexity, developing accurate and explainable detection systems is crucial for robust cybersecurity. This paper introduces a hybrid malware classification system that leverages fine-tuned large language models (LLMs) and an enriched cybersecurity knowledge graph (KG) to enhance both detection accuracy and interpretability. The system processes two types of input data—network packets and memory dumps, and classifies applications as benign or malign. Fine-tuned Llama models provide initial classifications, along with reasoning, which are further refined through prompt chaining using a knowledge graph populated with MITRE ATT&CK data and SecureBERT embeddings. The KG facilitates contextual reasoning, resulting in more accurate detection and informed decision-making. Experimental results demonstrate classification accuracies of 91.2% for network packet data and 94.35% for memory dump data, despite challenges related to LLM hallucinations and output parsing.

This paper investigates the persistent influence of historical redlining on modern AI algorithms used in real estate and loan approvals. Utilizing Home Mortgage Disclosure Act (HMDA) data, we uncover demographic biases in loan approval processes and track their evolution over time. Through the application of machine learning models and the bias detection and mitigation toolkit, we assess fairness using metrics such as statistical parity difference, disparate impact, and the Theil Index. Our analysis demonstrates the existence of discrimination, and shows that mitigation techniques, such as reweighting and domain knowledge inclusion, can significantly reduce disparities and promote equity in loan approvals across race, gender, and ethnicity. This study also highlights the necessity of addressing historical biases in training data to foster fairer algorithmic decision-making, while proposing practical solutions for improving fairness in AI-based lending systems.

The rate of terror attacks has surged over the past decade, resulting in the tragic and senseless loss or alteration of numerous lives. Offenders behind mass shootings, bombings, or other domestic terrorism incidents have historically exhibited warning signs on social media before carrying out actual incidents. However, due to inadequate and comprehensive police procedures, authorities and social media platforms are often unable to detect these early indicators of intent. To tackle this issue, we aim to create a multimodal model capable of predicting sentiments simultaneously from both images (i.e., social media photos) and text (i.e., social media posts), generating a unified prediction. The proposed method involves segregating the image and text components of an online post and utilizing a captioning model to generate sentences summarizing the image’s contents. Subsequently, a sentiment analyzer evaluates this caption, or description, along with the original post’s text to determine whether the post is positive (i.e., concerning) or negative (i.e., benign). This undertaking represents a significant step toward implementing the developed system in real-world scenarios.

Approximately 30% of all traffic fatalities in the United States are attributed to alcohol-impaired driving. This means that, despite stringent laws against this offense in every state, the frequency of drunk driving accidents is alarming, resulting in approximately one person being killed every 45 minutes. The process of charging individuals with Driving Under the Influence (DUI) is intricate and can sometimes be subjective, involving multiple stages such as observing the vehicle in motion, interacting with the driver, and conducting Standardized Field Sobriety Tests (SFSTs). Biases have been observed through racial profiling, leading to some groups and geographical areas facing fewer DUI tests, resulting in many actual DUI incidents going undetected, ultimately leading to a higher number of fatalities. To tackle this issue, our research introduces an Artificial Intelligence-based predictor that is both fairness-aware and incorporates domain knowledge to analyze DUI-related fatalities in different geographic locations. Through this model, we gain intriguing insights into the interplay between various demographic groups, including age, race, and income. By utilizing the provided information to allocate policing resources in a more equitable and efficient manner, there is potential to reduce DUI-related fatalities and have a significant impact on road safety.

The COVID-19 pandemic has affected millions of people worldwide with severe health, economic, social, and political implications. Healthcare Policy Makers (HPMs) and medical experts are at the core of responding to this continuously evolving pandemic situation and are working hard to contain the spread and severity of this relatively unknown virus. Biomedical researchers are continually discovering new information about this virus and communicating the findings through scientific articles. As such, it is crucial for HPMs and funding agencies to monitor the COVID-19 research trend globally on a regular basis. However, given the influx of biomedical research articles, monitoring COVID-19 research trends has become more challenging than ever, especially when HPMs want on-demand guided search techniques with a set of topics of interest in mind. Unfortunately, existing topic trend modeling techniques are unable to serve this purpose as (1) traditional topic models are unsupervised, and (2) HPMs in different regions may have different topics of interest that they want to track. To address this problem, we introduce a novel computational task in this article calledAd-Hoc Topic Tracking, which is essentially a combination ofzero-shottopic categorization and the spatio-temporal analysis task. We then propose multiplezero-shotclassification methods to solve this task by building on state-of-the-art language understanding techniques. Next, we picked the best-performing method based on its accuracy on a separate validation dataset and then applied it to a corpus of recent biomedical research articles to track COVID-19 research endeavors across the globe using a spatio-temporal analysis. A demo website has also been developed for HPMs to create custom spatio-temporal visualizations of COVID-19 research trends. The research outcomes demonstrate that the proposedzero-shotclassification methods can potentially facilitate further research on this important subject matter. At the same time, the spatio-temporal visualization tool will greatly assist HPMs and funding agencies in making well-informed policy decisions for advancing scientific research efforts.

Non-pharmaceutical Interventions (NPIs), such as Stay-at-Home, and Face-Mask-Mandate, are essential components of the public health response to contain an outbreak like COVID-19. However, it is very challenging to quantify the individual or joint effectiveness of NPIs and their impact on people from different racial and ethnic groups or communities in general. Therefore, in this paper, we study the following two research questions: 1) How can we quantitatively estimate the effectiveness of different NPI policies pertaining to the COVID-19 pandemic?; and 2) Do these policies have considerably different effects on communities from different races and ethnicity? To answer these questions, we model the impact of an NPI as a joint function of stringency and effectiveness over a duration of time. Consequently, we propose a novel stringency function that can provide an estimate of how strictly an NPI was implemented on a particular day. Next, we applied two popular tree-based discriminative classifiers, considering the change in daily COVID cases and death counts as binary target variables, while using stringency values of different policies as independent features. Finally, we interpreted the learned feature weights as the effectiveness of COVID-19 NPIs. Our experimental results suggest that, at the country level, restaurant closures and stay-at-home policies were most effective in restricting the COVID-19 confirmed cases and death cases respectively; and overall, restaurant closing was most effective in hold-down of COVID-19 cases at individual community levels such as Asian, White, Black, AIAN and, NHPI. Additionally, we also performed a comparative analysis between race-specific effectiveness and country-level effectiveness to see whether different communities were impacted differently. Our findings suggest that the different policies impacted communities (race and ethnicity) differently.

Efficient and scalable matrix operations are being highly demanding in the recent era of Machine Learning, Deep Learning, and Big Data Analytics. The two commonly used matrix-matrix operations in the Basic Linear Algebra Subprograms (BLAS) specification are General Matrix-Matrix multiplication (GEMM) and Symmetric Rank-k update (SYRK). The SYRK routine is a specialization of the GEMM routine, where half of the multiplications are skipped as the resultant matrix is known to be symmetric. Fortunately, several linear algebra libraries implement these BLAS routines quite efficiently. The libraries usually partition the input matrices into blocks and place them in processor caches, thus improving performance by leveraging the caches. However, the contemporary libraries are highly optimized for squarish matrices, but the performance degrades significantly for the matrices with edge case (strictly thin or strictly fat shapes) in the multicore machine. The primary reason is that the current state-of-the-art libraries make fixed block shapes based on a processor architecture, and do not consider the shape of the input matrices. In this paper, we propose a new blocking approach, we name it Flexible-blocking, to mitigate the scalability issues. In contrast to the contemporary libraries, our approach formulates the blocks of the input matrices based on the shapes of the matrices as well as the number of threads used in the implementation. Our proposed technique shows noticeable performance improvement on multicore shared-memory machines for the edge case matrices.

Illegal insider trading of stocks is based on releasing non-public information (e.g., new product launch, quarterly financial report, acquisition or merger plan) before the information is made public. Detecting illegal insider trading is difficult due to the complex, nonlinear, and non-stationary nature of the stock market. In this work, we present an approach that detects and predicts illegal insider trading proactively from large heterogeneous sources of structured and unstructured data using a deep-learning based approach combined with discrete signal processing on the time series data. In addition, we use a tree-based approach that visualizes events and actions to aid analysts in their understanding of large amounts of unstructured data. Using existing data, we have discovered that our approach has a good success rate in detecting illegal insider trading patterns.