The 0-day attack is a cyber-attack based on vulnerabilities that have not yet been published. The detection of anomalous traffic generated by such attacks is vital, as it can represent a critical problem, both in a technical and economic sense, for a smart enterprise as for any system largely dependent on technology. To predict this kind of attack, one solution can be to use unsupervised machine learning approaches, as they guarantee the detection of anomalies regardless of their prior knowledge. It is also essential to identify the anomalous and unknown behaviors that occur within a network in near real-time. Three different approaches have been proposed and benchmarked in exactly the same condition: Deep Autoencoding with GMM and Isolation Forest, Deep Autoencoder with Isolation Forest, and Memory Augmented Deep Autoencoder with Isolation Forest. These approaches are thus the result of combining different unsupervised algorithms. The results show that the addition of the Isolation Forest improves the accuracy values and increases the inference time, although this increase does not represent a relevant problematic factor. This paper also explains the features that the various models consider most important for classifying an event as an attack using the explainable artificial intelligence methodology called Shapley Additive Explanations (SHAP). Experiments were conducted on KDD99, NSL-KDD, and CIC-IDS2017 datasets.

Web phishing is a form of cybercrime aimed at tricking people into visiting malicious URLs to exfiltrate sensitive data. Since the structure of a malicious URL evolves over time, phishing detection mechanisms that can adapt to such variations are paramount. Furthermore, web phishing detection is an unbalanced classification task, as legitimate URLs outnumber malicious ones in real-life cases. Deep learning (DL) has emerged as a promising technique to minimize concept drift to enhance web phishing detection. Deep reinforcement learning (DRL) combines DL with reinforcement learning (RL); that is, a sequential decision-making paradigm in which the problem to be addressed is expressed as a Markov decision process (MDP). Recent studies have proposed an ad hoc MDP formulation to tackle unbalanced classification tasks called the imbalanced classification Markov decision process (ICMDP). In this paper, we exploit the ICMDP to present a double deep Q-Network (DDQN)-based classifier to address the unbalanced web phishing classification problem. The proposed algorithm is evaluated on a Mendeley web phishing dataset, from which three different data imbalance scenarios are generated. Despite a significant training time, it results in better geometric mean, index of balanced accuracy, F1 score, and area under the ROC curve than other DL-based classifiers combined with data-level sampling techniques in all test cases.

Recognition of malware is critical in cybersecurity as it allows for avoiding execution and the downloading of malware. One of the possible approaches is to analyze the executable’s Application Programming Interface (API) calls, which can be done using tools that work in sandboxes, such as Cuckoo or CAPEv2. This chain of calls can then be used to classify if the considered file is benign or malware. This work aims to compare six modern shallow learning and deep learning techniques based on tabular data, using two datasets of API calls containing malware and goodware, where the corresponding chain of API calls is expressed for each instance. The results show the quality of shallow learning approaches based on tree ensembles, such as CatBoost, both in terms of F1-macro score and Area Under the ROC curve (AUC ROC), and training time, making them optimal for making inferences on Edge AI solutions. The results are then analyzed with the explainable AI SHAP technique, identifying the API calls that most influence the process, i.e., those that are particularly afferent to malware and goodware.

Reports produced by popular malware analysis services showed a disparity in samples available for different malware families. The unequal distribution between such classes can be attributed to several factors, such as technological advances and the application domain that seeks to infect a computer virus. Recent studies have demonstrated the effectiveness of deep learning (DL) algorithms when learning multi-class classification tasks using imbalanced datasets. This can be achieved by updating the learning function such that correct and incorrect predictions performed on the minority class are more rewarded or penalized, respectively. This procedure can be logically implemented by leveraging the deep reinforcement learning (DRL) paradigm through a proper formulation of the Markov decision process (MDP). This paper proposes SINNER, i.e., a DRL-based multi-class classifier that approaches the data imbalance problem at the algorithmic level by exploiting a redesigned reward function, which modifies the traditional MDP model used to learn this task. Based on the experimental results, the proposed formula appears to be successful. In addition, SINNER has been compared to several DL-based models that can handle class skew without relying on data-level techniques. Using three out of four datasets sourced from the existing literature, the proposed model achieved state-of-the-art classification performance.

As the use of information systems exponentially increases, every organization is exposed to cyber-attacks. To detect and mitigate the damage caused by such attacks, organizations need to share information extracted from the analysis of known ones. Intrusion detection (IDS) and intrusion prevention (IPS) systems use information on known threats to detect and prevent attack re-execution. Given the large amount of information usually available for known attacks, restricting such amount only to really valuable information is necessary to allow protection systems to work more efficiently. One of the main challenges in cyber threat intelligence (CTI) is to filter relevant information and eliminate obsolete data. The recently published RFC 9424 emphasizes the need to produce such systems. In this work, a methodology named comprehensive assessment and rating of IoCs via CADE algorithm (CARIOCA) is proposed, which aims to analyze data contained in the CTI platform to select a subset of indicators of compromise (IoCs) considered most relevant for protection systems. Through CARIOCA, IoCs evaluation based on three level scorings is proposed, considering sources’ reliability, IoCs freshness, and CTI reports quality using a new algorithm, named category attribute density evaluation (CADE). The state-of-the-art considers the qualities of an IoC or the estimated reliability of the CTI source to select relevant IoCs. By combining three scores, CARIOCA can comprehensively assess IoCs relevance. The results obtained in the experiments support CARIOCA’s effectiveness in selecting the most relevant subset of IoCs for IDS/IPS.