In machine learning (ML), a widespread adage is that the area under the precision-recall curve (AUPRC) is a superior metric for model comparison to the area under the receiver operating characteristic (AUROC) for binary classification tasks with class imbalance. This paper challenges this notion through novel mathematical analysis, illustrating that AUROC and AUPRC can be concisely related in probabilistic terms. We demonstrate that AUPRC, contrary to popular belief, is not superior in cases of class imbalance and might even be a harmful metric, given its inclination to unduly favor model improvements in subpopulations with more frequent positive labels. This bias can inadvertently heighten algorithmic disparities. Prompted by these insights, a thorough review of existing ML literature was conducted, utilizing large language models to analyze over 1.5 million papers from arXiv. Our investigation focused on the prevalence and substantiation of the purported AUPRC superiority. The results expose a significant deficit in empirical backing and a trend of misattributions that have fuelled the widespread acceptance of AUPRC's supposed advantages. Our findings represent a dual contribution: a significant technical advancement in understanding metric behaviors and a stark warning about unchecked assumptions in the ML community. All experiments are accessible at https://github.com/mmcdermott/AUC_is_all_you_need.

Language models pre-trained on web-scale corpora demonstrate impressive capabilities on diverse downstream tasks. However, there is increasing concern whether such capabilities might arise from evaluation datasets being included in the pre-training corpus -- a phenomenon known as \textit{data contamination} -- in a manner that artificially increases performance. There has been little understanding of how this potential contamination might influence LMs' performance on downstream tasks. In this paper, we explore the impact of data contamination at the pre-training stage by pre-training a series of GPT-2 models \textit{from scratch}. We highlight the effect of both text contamination (\textit{i.e.}\ input text of the evaluation samples) and ground-truth contamination (\textit{i.e.}\ the prompts asked on the input and the desired outputs) from evaluation data. We also investigate the effects of repeating contamination for various downstream tasks. Additionally, we examine the prevailing n-gram-based definitions of contamination within current LLM reports, pinpointing their limitations and inadequacy. Our findings offer new insights into data contamination's effects on language model capabilities and underscore the need for independent, comprehensive contamination assessments in LLM studies.

Humans have good natural intuition to recognize when another person has something to say. It would be interesting if an AI can also recognize intentions to speak. Especially in scenarios when an AI is guiding a group discussion, this can be a useful skill. This work studies the inference of successful and unsuccessful intentions to speak from accelerometer data. This is chosen because it is privacy-preserving and feasible for in-the-wild settings since it can be placed in a smart badge. Data from a real-life social networking event is used to train a machine-learning model that aims to infer intentions to speak. A subset of unsuccessful intention-to-speak cases in the data is annotated. The model is trained on the successful intentions to speak and evaluated on both the successful and unsuccessful cases. In conclusion, there is useful information in accelerometer data, but not enough to reliably capture intentions to speak. For example, posture shifts are correlated with intentions to speak, but people also often shift posture without having an intention to speak, or have an intention to speak without shifting their posture. More modalities are likely needed to reliably infer intentions to speak.

The detection of anomalies in multivariate time series data is crucial for various practical applications, including smart power grids, traffic flow forecasting, and industrial process control. However, real-world time series data is usually not well-structured, posting significant challenges to existing approaches: (1) The existence of missing values in multivariate time series data along variable and time dimensions hinders the effective modeling of interwoven spatial and temporal dependencies, resulting in important patterns being overlooked during model training; (2) Anomaly scoring with irregularly-sampled observations is less explored, making it difficult to use existing detectors for multivariate series without fully-observed values. In this work, we introduce a novel framework called GST-Pro, which utilizes a graph spatiotemporal process and anomaly scorer to tackle the aforementioned challenges in detecting anomalies on irregularly-sampled multivariate time series. Our approach comprises two main components. First, we propose a graph spatiotemporal process based on neural controlled differential equations. This process enables effective modeling of multivariate time series from both spatial and temporal perspectives, even when the data contains missing values. Second, we present a novel distribution-based anomaly scoring mechanism that alleviates the reliance on complete uniform observations. By analyzing the predictions of the graph spatiotemporal process, our approach allows anomalies to be easily detected. Our experimental results show that the GST-Pro method can effectively detect anomalies in time series data and outperforms state-of-the-art methods, regardless of whether there are missing values present in the data. Our code is available: https://github.com/huankoh/GST-Pro.

Traumatic Brain Injury (TBI) poses a significant global public health challenge, contributing to high morbidity and mortality rates and placing a substantial economic burden on healthcare systems worldwide. The diagnosis of TBI relies on clinical information along with Computed Tomography (CT) scans. Addressing the multifaceted challenges posed by TBI has seen the development of innovative, data-driven approaches, for this complex condition. Particularly noteworthy is the prevalence of mild TBI (mTBI), which constitutes the majority of TBI cases where conventional methods often fall short. As such, we review the state-of-the-art Machine Learning (ML) techniques applied to clinical information and CT scans in TBI, with a particular focus on mTBI. We categorize ML applications based on their data sources, and there is a spectrum of ML techniques used to date. Most of these techniques have primarily focused on diagnosis, with relatively few attempts at predicting the prognosis. This review may serve as a source of inspiration for future research studies aimed at improving the diagnosis of TBI using data-driven approaches and standard diagnostic data.

Uncertainty estimation in Neural Networks (NNs) is vital in improving reliability and confidence in predictions, particularly in safety-critical applications. Bayesian Neural Networks (BayNNs) with Dropout as an approximation offer a systematic approach to quantifying uncertainty, but they inherently suffer from high hardware overhead in terms of power, memory, and computation. Thus, the applicability of BayNNs to edge devices with limited resources or to high-performance applications is challenging. Some of the inherent costs of BayNNs can be reduced by accelerating them in hardware on a Computation-In-Memory (CIM) architecture with spintronic memories and binarizing their parameters. However, numerous stochastic units are required to implement conventional dropout-based BayNN. In this paper, we propose the Scale Dropout, a novel regularization technique for Binary Neural Networks (BNNs), and Monte Carlo-Scale Dropout (MC-Scale Dropout)-based BayNNs for efficient uncertainty estimation. Our approach requires only one stochastic unit for the entire model, irrespective of the model size, leading to a highly scalable Bayesian NN. Furthermore, we introduce a novel Spintronic memory-based CIM architecture for the proposed BayNN that achieves more than $100\times$ energy savings compared to the state-of-the-art. We validated our method to show up to a $1\%$ improvement in predictive performance and superior uncertainty estimates compared to related works.

Online changepoint detection aims to detect anomalies and changes in real-time in high-frequency data streams, sometimes with limited available computational resources. This is an important task that is rooted in many real-world applications, including and not limited to cybersecurity, medicine and astrophysics. While fast and efficient online algorithms have been recently introduced, these rely on parametric assumptions which are often violated in practical applications. Motivated by data streams from the telecommunications sector, we build a flexible nonparametric approach to detect a change in the distribution of a sequence. Our procedure, NP-FOCuS, builds a sequential likelihood ratio test for a change in a set of points of the empirical cumulative density function of our data. This is achieved by keeping track of the number of observations above or below those points. Thanks to functional pruning ideas, NP-FOCuS has a computational cost that is log-linear in the number of observations and is suitable for high-frequency data streams. In terms of detection power, NP-FOCuS is seen to outperform current nonparametric online changepoint techniques in a variety of settings. We demonstrate the utility of the procedure on both simulated and real data.

A key challenge in the continual learning setting is to efficiently learn a sequence of tasks without forgetting how to perform previously learned tasks. Many existing approaches to this problem work by either retraining the model on previous tasks or by expanding the model to accommodate new tasks. However, these approaches typically suffer from increased storage and computational requirements, a problem that is worsened in the case of sparse models due to need for expensive re-training after sparsification. To address this challenge, we propose a new method for efficient continual learning of sparse models (EsaCL) that can automatically prune redundant parameters without adversely impacting the model's predictive power, and circumvent the need of retraining. We conduct a theoretical analysis of loss landscapes with parameter pruning, and design a directional pruning (SDP) strategy that is informed by the sharpness of the loss function with respect to the model parameters. SDP ensures model with minimal loss of predictive accuracy, accelerating the learning of sparse models at each stage. To accelerate model update, we introduce an intelligent data selection (IDS) strategy that can identify critical instances for estimating loss landscape, yielding substantially improved data efficiency. The results of our experiments show that EsaCL achieves performance that is competitive with the state-of-the-art methods on three continual learning benchmarks, while using substantially reduced memory and computational resources.

Self-supervised learning has become a popular way to pretrain a deep learning model and then transfer it to perform downstream tasks. However, most of these methods are developed on large-scale image datasets that contain natural objects with clear textures, outlines, and distinct color contrasts. It remains uncertain whether these methods are equally effective for medical imaging, where the regions of interest often blend subtly and indistinctly with the surrounding tissues. In this study, we propose an alternative method that uses contralateral mammograms to train a neural network to encode similar embeddings when a pair contains both normal images and different embeddings when a pair contains normal and abnormal images. Our approach leverages the natural symmetry of human body as weak labels to learn to distinguish abnormal lesions from background tissues in a fully unsupervised manner. Our findings suggest that it's feasible by incorporating soft labels derived from the Euclidean distances between the embeddings of the image pairs into the Siamese network loss. Our method demonstrates superior performance in mammogram patch classification compared to existing self-supervised learning methods.

Social engineering (SE) aims at deceiving users into performing actions that may compromise their security and privacy. These threats exploit weaknesses in human's decision making processes by using tactics such as pretext, baiting, impersonation, etc. On the web, SE attacks include attack classes such as scareware, tech support scams, survey scams, sweepstakes, etc., which can result in sensitive data leaks, malware infections, and monetary loss. For instance, US consumers lose billions of dollars annually due to various SE attacks. Unfortunately, generic social engineering attacks remain understudied, compared to other important threats, such as software vulnerabilities and exploitation, network intrusions, malicious software, and phishing. The few existing technical studies that focus on social engineering are limited in scope and mostly focus on measurements rather than developing a generic defense. To fill this gap, we present SEShield, a framework for in-browser detection of social engineering attacks. SEShield consists of three main components: (i) a custom security crawler, called SECrawler, that is dedicated to scouting the web to collect examples of in-the-wild SE attacks; (ii) SENet, a deep learning-based image classifier trained on data collected by SECrawler that aims to detect the often glaring visual traits of SE attack pages; and (iii) SEGuard, a proof-of-concept extension that embeds SENet into the web browser and enables real-time SE attack detection. We perform an extensive evaluation of our system and show that SENet is able to detect new instances of SE attacks with a detection rate of up to 99.6% at 1% false positive, thus providing an effective first defense against SE attacks on the web.