As the number of intelligent systems has dramatically increased, malware has become increasingly harmful. Consequently, detecting previously unknown malware swiftly has emerged as a critical issue in cybersecurity, aiming to prevent these threats from becoming widespread hazards that harm users and systems. To address this challenge, machine learning methods have been employed to rapidly detect and classify malware. These methods can be either dynamic or static analysis. Dynamic analysis allows for the collection of significant features not obtainable through static analysis alone. However, malware authors often employ techniques to evade automated malware analysis, leading to decreased true detection rates due to sandbox evasion strategies. In this study we focus on static analysis for malware detection, specifically on Portable Executable (PE) files. For this purpose, we propose a novel feature extraction pipeline, and a methodology aimed at finding reliably secure detectors, where the main priority is to avoid at all cost false negatives. For this, we leverage the use of the AutoML (auto-sklearn) framework and perform a reliability analysis of different configurations after consequently setting the working points to avoid false negatives. Our experiments, performed using the up-to-date MalwareBazaar malware dataset, show that it is possible to find consistently robust ML detectors when using the proposed static feature sets for PE files.

Geofencing, the virtual demarcation of physical spaces, is widely used for managing the localisation of Internet of Things (IoT) devices. However, traditional localisation techniques face security challenges indoors due to signal interference and susceptibility to spoofing, often requiring extensive calibration or extra hardware, limiting scalability. In this work, we propose ZeroTouch, a machine learning-based system that leverages Received Signal Strength (RSS) measurements from multiple receivers to improve the security of geofencing without introducing additional deployment overhead. While RSS-based localisation is known to have inherent security limitations, we show that by aggregating RSS readings from multiple anchor points and detecting anomalies using an autoencoder model, ZeroTouch provides a practical and automated mechanism for verifying whether a device is inside or outside a defined boundary. Rather than serving as a standalone security mechanism, ZeroTouch enhances existing authentication frameworks by adding an additional zero-touch security layer that operates passively in the background. ZeroTouch eliminates manual calibration, removes the human-in-the-loop element, and simplifies deployment. We evaluate our solution in a realistic simulated environment and demonstrate that it achieves high accuracy in distinguishing between in-room and out-of-room devices, even in strong adversarial settings.

Biometric authentication systems have traditionally relied on the False Match Rate (FMR) to evaluate security against impersonation threats. However, this metric alone is insufficient for assessing vulnerabilities to statistical attacks because it cannot account for the non-uniformity of mismatches and atypical inputs that adversaries may manipulate. To address this issue, we propose a new evaluation framework that overcomes these limitations. The framework includes an estimate of the effective key space of biometrics and metrics that consider non-uniformity in the biometric embedding space. Our findings demonstrate that our framework provides a nuanced understanding of biometric security. Moreover, optimizing for the proposed metric leads to better security against statistical attacks than optimizing the FMR. Furthermore, the framework provides a comparative security analysis with traditional methods like passwords and PIN codes. It also quantifies the impact on security when adversaries partially know their victims, e.g., demographics.

Biometric template protection is a key technology for the successful integration of biometric recognition in real-world use cases. However, single-factor biometric solutions have their own drawbacks, and even when a second factor is used, its disclosure can result in important information leakages about the biometric template, with a serious impact on security and privacy. In this paper, we introduce a multifactor Strengthened Fuzzy Extractor (SFEs) cryptosystem using efficient Turbo-Codes as Error Correcting Code and apply it to finger veins as a high-performance biometric trait. In our experiments we show the effectiveness of using user-specific helper data, by comparing two user-dependent approaches to the most common user-independent one. The use of a cryptosystem that takes into account the variability of the specific template, and a high-performance biometric trait, provides higher and more informed security. This contributes towards a more flexible and realistic integration of biometrics in less constrained scenarios.

An overwhelming number of alerts – especially false ones – can desensitize analysts in security operations centers (SOC), possibly resulting in missed critical incidents and attacks going unnoticed. With inadequate alert monitoring, improper thresholds, and missing feedback loops as lead causes of alert fatigue, we investigate the use of automated machine learning to increase the efficiency of a SOC through automation of false alerts filtering. More specifically, we design a methodology to allow a safer use of AutoML to reduce false alerts, and validate this on a real-world case study. To be more precise, our approach is tailored to address datasets that exhibit limited instances of true positives, possess high dimensionality relative to their size, and demonstrate temporal fluctuations. We have identified diverse setups that provide comparable and reliably effective results in minimizing false positive alerts, all the while avoiding instances of false negatives. Furthermore, we provide valuable insights into the application of these automated frameworks within the realm of security.

Stream-based biometric authentication using a novel approach based on spiking neural networks (SNNs) is addressed. SNNs have proven advantages regarding energy consumption and they are a perfect match with some proposed neuromorphic hardware chips, which can lead to a broader adoption of user device applications of artificial intelligence technologies. One of the challenges when using SNNs is the discriminative training of the network since it is not straightforward to apply the well-known error backpropagation (EBP), massively used in traditional artificial neural networks (ANNs). A network structure based on neuron columns is proposed, resembling cortical columns in the human cortex, and a new derivation of error backpropagation for the spiking neural networks that integrate the lateral inhibition in these structures. The potential of the proposed approach is tested in the task of inertial gait authentication, where gait is quantified as signals from Inertial Measurement Units (IMU), and the authors' approach to state-of-the-art ANNs is compared. In the experiments, SNNs provide competitive results, obtaining a difference of around 1% in half total error rate when compared to state-of-the-art ANNs in the context of IMU-based gait authentication.

As biometric templates consist of highly correlated features, the real security level offered by biometric authentication systems remains an open research question. In this work we provide new approximations and a lower bound of the security offered by fuzzy commitment schemes. Fuzzy commitment cryptosystems and in general biometric template protection schemes play an important role in allowing for remote storage and processing of biometric data, as they mitigate the threat of biometric template leakage. The use of such schemes would alleviate some of the usability constraints imposed by the state-of-practice local use of biometrics. As such we conduct an in-depth security analysis for IMU based gait authentication systems, where we evaluate the effectiveness of attacks within the scope of two well-defined threat models that target both unprotected and protected systems. A pivotal enabler of our analysis is the development of nine different approaches to gait authentication, which allows us to perform intramodal fusion on these distinct, yet highly correlated biometric templates, and to protect them with a strengthened fuzzy commitment scheme. Our analysis clearly demonstrates the high correlation between the different biometric templates, which, among others, further showcases the threat of biometric template leakage. Furthermore, as our analysis incorporates a threat model that assumes biometric template leakage, it provides metrics for the security provided by the biometric modality itself.

Sensors provide the foundation of many smart applications and cyber–physical systems by measuring and processing information upon which applications can make intelligent decisions or inform their users. Inertial measurement unit (IMU) sensors—and accelerometers and gyroscopes in particular—are readily available on contemporary smartphones and wearable devices. They have been widely adopted in the area of activity recognition, with fall detection and step counting applications being prominent examples in this field. However, these sensors may also incidentally reveal sensitive information in a way that is not easily envisioned upfront by developers. Far worse, the leakage of sensitive information to third parties, such as recommender systems or targeted advertising applications, may cause privacy concerns for unsuspecting end-users. In this paper, we explore the elicitation of age and gender information from gait traces obtained from IMU sensors, and systematically compare different feature engineering and machine learning algorithms, including both traditional and deep learning methods. We describe in detail the prediction methods that our team used in the OU-ISIR Wearable Sensor-based Gait Challenge: Age and Gender (GAG 2019) at the 12th IAPR International Conference on Biometrics. In these two competitions, our team obtained the best solutions amongst all international participants, and this for both the age and gender predictions. Our research shows that it is feasible to predict age and gender with a reasonable accuracy on gait traces of just a few seconds. Furthermore, it illustrates the need to put in place adequate measures in order to mitigate unintended information leakage by abusing sensors as an unanticipated side channel for sensitive information or private traits.

With the increasing capabilities of wearable sensors and implantable medical devices, new opportunities arise to diagnose, control and treat several chronic conditions. Unfortunately, these advancements also open new attack vectors, making security an essential requirement for the further adoption of these devices. Researchers have already developed security solutions tailored to their unique requirements and constraints. However, a fundamental yet unsolved problem is how to securely and efficiently establish and manage cryptographic keys. One of the most promising approaches is the use of patient's physiological signals for key establishment. This paper aims at identifying common pitfalls in physiological-signal-based cryptographic protocols. These solutions are very fragile because errors can be introduced at different stages, including the choice of the physiological signal, the design of the protocol or its implementation. We start by reviewing previous work that has succeeded in measuring various physiological signals remotely. Subsequently, we conduct a thorough security analysis of two cryptographic solutions well-accepted by the security community, namely the H2H protocol (Rostami et al. - CCS 2013) and the Biosec protocol (Cherukuri et al. - ICISIP 2006). Our evaluation reveals that these protocols have serious design and implementation security weaknesses. Driven by our findings, we then describe how to use fuzzy extractors for designing secure and efficient cryptographic solutions based on the patients' physiological signals. Finally, we discuss research directions for future work.

In this paper, we show that a recently published lightweight adaptation of a Fingerprint matching algorithm called the Minutia Cylinder-Code may not be secure as intruders may be able to illegitimately yet successfully authenticate themselves to the system under consideration. We also show that the lightweight adaptation has other privacy related vulnerabilities that make it unsuitable for use in Biometrics. We make it clear that we are neither investigating nor commenting on the security of the original Minutia Cylinder-Code algorithm by itself, rather we highlight the vulnerabilities of the lightweight adaptation. In the process of doing this, we provide a high-level overview of the role of one-way functions in cryptography and biometrics to provide a context to the aforementioned lightweight algorithm and its deficiencies.

The procedure for extracting a cryptographic key from noisy sources, such as biometrics and physically uncloneable functions (PUFs), is known as fuzzy extractor (FE). Although FE constructions deal with discrete sources, most noisy sources are continuous. In the continuous case, it is required to transform the source to a discrete one. We introduce a 1) model-based uncoupling construction that directly deals with the continuous noisy source and produces helper data uncoupling the discrete representation from the noisy source, guaranteeing the diversity of the discrete representation, and making it more robust and a 2) strengthened uncoupled fuzzy extractor, suitable for privacy-preserving applications, which integrates an additional fixed authentication factor and obtains a key uncoupled to the noisy sources and unlinkable helper data. We present optimal model-based uncoupling constructions for Gaussian sources. Specifically, we show how to: 1) extract one or multiple bits from a single Gaussian source; 2) extract one bit from several unreliable Gaussian sources; and 3) provide a general procedure to obtain an optimal uncoupled FE from Gaussian source(s). Our experiments show that the proposed constructions achieve much higher security levels for wide operational scenarios, approximately doubling the obtained effective key length without affecting false rejection rates.

Authentication and authorization are critical security layers to protect a wide range of online systems, services and content. However, the increased prevalence of wearable and mobile devices, the expectations of a frictionless experience and the diverse user environments will challenge the way users are authenticated. Consumers demand secure and privacy-aware access from any device, whenever and wherever they are, without any obstacles. This paper reviews emerging trends and challenges with frictionless authentication systems and identifies opportunities for further research related to the enrollment of users, the usability of authentication schemes, as well as security and privacy trade-offs of mobile and wearable continuous authentication systems.

This paper presents a hidden Markov model-Universal background model gait authentication system, which is also incorporated into a template protection based on a fuzzy commitment scheme. We show that with limited enrollment data the HMM-UBM system achieves a very competitive equal error rate of ≈1% using one sensor. The proposed template protection scheme benefits from eigen-features coming from multiple Universal background model systems fused with a novel technique that minimizes the bit error rate for genuine attempts. This allows the protected system to achieve a false rejection rate below 5% with an effective key length of 64 bits.

This paper proposes a frictionless authentication system, provides a comprehensive security analysis of and proposes potential solutions for this system. It first presents a system that allows users to authenticate to services in a frictionless manner, i.e., without the need to perform any particular authentication-related actions. Based on this system model, the paper analyses security problems and potential privacy threats imposed on users, leading to the specification of a set of security and privacy requirements. These requirements can be used as a guidance on designing secure and privacy-friendly frictionless authentication systems. The paper also sketches three potential solutions for such systems and highlights their advantages and disadvantages.

Biometrics are widely used for authentication in several domains, services and applications. However, only very few systems succeed in effectively combining highly secure user authentication with an adequate privacy protection of the biometric templates, due to the difficulty associated with jointly providing good authentication performance, unlinkability and irreversibility to biometric templates. This thwarts the use of biometrics in remote authentication scenarios, despite the advantages that this kind of architectures provides. We propose a user-specific approach for decoupling the biometrics from their binary representation before using biometric protection schemes based on fuzzy extractors. This allows for more reliable, flexible, irreversible and unlinkable protected biometric templates. With the proposed biometrics decoupling procedures, biometric metadata, that does not allow to recover the original biometric template, is generated. However, different biometric metadata that are generated starting from the same biometric template remain statistically linkable, therefore we propose to additionally protect these using a second authentication factor (e.g., knowledge or possession based). We demonstrate the potential of this approach within a two-factor authentication protocol for remote biometric authentication in mobile scenarios.

User authentication based on biometrics is getting an increasing attention. However, privacy concerns for biometric data have impeded the adoption of cloud-based services for biometric authentication. This paper proposes an efficient distributed two-factor authentication protocol that is privacy-preserving even in the presence of colluding internal adversaries. One of the authentication factors in our protocol is biometrics, and the other factor can be either knowledge-based or possession-based. The actors involved in our protocol are users, user/client devices with biometric sensors, service provider, and cloud for storing protected biometric templates. Contrary to the existing biometric authentication protocols that offer security only in the honest-but-curious adversarial model, our protocol provides enhanced security and privacy properties in the active (or malicious) adversarial model. Specifically, our protocol offers identity privacy, unlinkability, and user data (i.e., the biometric template data and the second factor) privacy against compromised cloud storage service, and preserves the privacy of the user data even if the cloud storage service colludes with the service provider. Moreover, our protocol only employs lightweight schemes and thus is efficient. The distributed model combined with the security and privacy properties of our protocol paves the way towards a new cloud-based business model for privacy-preserving authentication.

This work addresses the security and privacy issues in remote biometric authentication by proposing an efficient mechanism to verify the correctness of the outsourced computation in such protocols. In particular, we propose an efficient verifiable computation of XORing encrypted messages using an XOR linear message authentication code (MAC) and we employ the proposed scheme to build a biometric authentication protocol. The proposed authentication protocol is both secure and privacy-preserving against malicious (as opposed to honest-but-curious) adversaries. Specifically, the use of the verifiable computation scheme together with an homomorphic encryption protects the privacy of biometric templates against malicious adversaries. Furthermore, in order to achieve unlinkability of authentication attempts, while keeping a low communication overhead, we show how to apply Oblivious RAM and biohashing to our protocol. We also provide a proof of security for the proposed solution. Our simulation results show that the proposed authentication protocol is efficient.

Implantable Medical Devices (IMDs) are used to monitor and control patients with chronic diseases. A growing number of IMDs are equipped with a wireless interface that allows non-invasive monitoring and reprogramming through an external device, also known as device programmer. However, this wireless interface also brings important security and privacy risks that may lead to remote attacks. In this domain, the use of cryptography is challenging due to the inherent tensions between security vs. accessibility and security vs. energy cost. A well-studied problem yet unsolved is how to establish (and manage) cryptographic keys between the device programmer and the IMD. Recent work has investigated how Physiological Signals (PS) extracted from the patient can be used for key agreement or authentication between the devices. This paper surveys some of the proposed countermeasures in the field of medical device security, with a special focus on those that use patient’s physiological signals for key establishment or authentication between the devices. We point out that most of the existing solutions, including those relying on PS, take assumptions that do not necessarily hold in practical scenarios. Furthermore, we show that the H2H protocol and the Biosec protocol have serious security weaknesses and design flaws which make them vulnerable to attacks. Based on our analysis, we define some of the challenges that need be addressed before adopting these solutions. Furthermore, we investigate how to use physiological-signal-based protocols in cryptography, possibly in combination with other solutions, such as pre-installed factory keys, to achieve higher security protection.

Parameterized Appearance Models, such as Active Appearance Models (AAM), Morphable Models, or Boosted Appearance Models, have been extensively used for face alignment. Discriminative methods learn a mapping function between appearance features and shape parameters. Different mapping functions have been studied in the literature, including linear regression, which has proved to perform well when close to the true solution. Despite its easiness, it still suffers from two major drawbacks: 1) It takes the whole data without highlighting relations among different regions of the face, and 2) it is computationally expensive both in time and memory. In this paper, we analyze the covariance of the training data, and propose a way to find related information. By clustering those patches that are related, we reach a noise-reduced regression matrix. Then, we construct a clean mapping matrix, with reduced dimensionality, taking only the relevant training information. Experiments show that this method outperforms linear regression for face alignment.