In addition to signature-based and heuristics-based detection techniques, machine learning (ML) is widely used to generalize to new, never-before-seen malicious software (malware). However, it has been demonstrated that ML models can be fooled by tricking the classifier into returning the incorrect label. These studies, for instance, usually rely on a prediction score that is fragile to gradient-based attacks. In the context of a more realistic situation where an attacker has very little information about the outputs of a malware detection engine, modest evasion rates are achieved [1]. In this paper, we propose a method using reinforcement learning with DQN and REINFORCE algorithms to challenge two state-of-the-art ML-based detection engines (MalConv & EMBER) and a commercial antivirus (AV) classified by Gartner as a leader AV [2]. Our method combines several actions, modifying a Windows portable execution (PE) file without breaking its functionalities. Our method also identifies which actions perform better and compiles a detailed vulnerability report to help mitigate the evasion. We demonstrate that REINFORCE achieves very good evasion rates even on a commercial AV with limited available information.
Evaluating error that arises through the aggregation of data recorded by multiple observers is a key consideration in many metric and geometric morphometric analyses of stone tool shape. One of the most common approaches involves the convergence of observers for repeat trails on the same set of artefacts: however, this is logistically and financially challenging when collaborating internationally and/or at a large scale. We present and evaluate a unique alternative for testing inter-observer error, involving the development of 3D printed copies of a lithic reference collection for distribution among observers. With the aim of reducing error, clear protocols were developed for photographing and measuring the replicas, and inter-observer variability was assessed on the replicas in comparison with a corresponding data set recorded by a single observer. Our results demonstrate that, when the photography procedure is standardized and dimensions are clearly defined, the resulting metric and geometric morphometric data are minimally affected by inter-observer error, supporting this method as an effective solution for assessing error under collaborative research frameworks. Collaboration is becoming increasingly important within archaeological and anthropological sciences in order to increase the accessibility of samples, encourage dual-project development between foreign and local researchers and reduce the carbon footprint of collection-based research. This study offers a promising validation of a collaborative research design whereby researchers remotely work together to produce comparable data capturing lithic shape variability.
Purpose: To measure core temperature (Tcore) in open-water (OW) swimmers during a 25-km competition and identify the predictors of Tcore drop and hypothermia-related dropouts. Methods: Twenty-four national- and international-level OW swimmers participated in the study. Participants completed a personal questionnaire and a body fat/muscle mass assessment before the race. The average speed was calculated on each lap over a 2500-m course. Tcore was continuously recorded via an ingestible temperature sensor (e-Celsius, BodyCap). Hypothermia-related dropouts (H group) were compared with finishers (nH group). Results: Average prerace Tcore was 37.5°C (0.3°C) (N = 21). 7 participants dropped out due to hypothermia (H, n = 7) with a mean Tcore at dropout of 35.3°C (1.5°C). Multiple logistic regression analysis found that body fat percentage and initial Tcore were associated with hypothermia (G2 = 17.26, P < .001). Early Tcore drop ≤37.1°C at 2500 m was associated with a greater rate of hypothermia-related dropouts (71.4% vs 14.3%, P = .017). Multiple linear regression found that body fat percentage and previous participation were associated with Tcore drop (F = 4.95, P = .019). There was a positive correlation between the decrease in speed and Tcore drop (r = .462, P < .001). Conclusions: During an OW 25-km competition at 20°C to 21°C, lower initial Tcore and lower body fat, as well as premature Tcore drop, were associated with an increased risk of hypothermia-related dropout. Lower body fat and no previous participation, as well as decrease in swimming speed, were associated with Tcore drop.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.