Producing labels for unlabeled data is error-prone, making semi-supervised learning (SSL) troublesome. Often, little is known about when and why an algorithm fails to outperform a supervised baseline. Using benchmark datasets, we craft five common real-world SSL data scenarios: few-label, open-set, noisy-label, and class distribution imbalance/misalignment in the labeled and unlabeled sets. We propose a novel algorithm called Contrastive Credibility Propagation (CCP) for deep SSL via iterative transductive pseudo-label refinement. CCP unifies semi-supervised learning and noisy label learning for the goal of reliably outperforming a supervised baseline in any data scenario. Compared to prior methods which focus on a subset of scenarios, CCP uniquely outperforms the supervised baseline in all scenarios, supporting practitioners when the qualities of labeled or unlabeled data are unknown.

Community detection, which focuses on clustering vertex interactions, plays a significant role in network analysis. However, it also faces numerous challenges like missing data and adversarial attack. How to further improve the performance and robustness of community detection for real-world networks has raised great concerns. In this paper, we explore robust community detection by enhancing network structure, with two generic algorithms presented: one is named robust community detection via genetic algorithm (RobustECD-GA), in which the modularity and the number of clusters are combined in a fitness function to find the optimal structure enhancement scheme; the other is called robust community detection via similarity ensemble (RobustECD-SE), integrating multiple information of community structures captured by various vertex similarities, which scales well on large-scale networks. Comprehensive experiments on real-world networks demonstrate, by comparing with two traditional enhancement strategies, that the new methods help six representative community detection algorithms achieve more significant performance improvement. Moreover, experiments on the corresponding adversarial networks indicate that the new methods could also optimize the network structure to a certain extent, achieving stronger robustness against adversarial attack.

In network link prediction, it is possible to hide a target link from being predicted with a small perturbation on network structure. This observation may be exploited in many real world scenarios, for example, to preserve privacy, or to exploit financial security. There have been many recent studies to generate adversarial examples to mislead deep learning models on graph data. However, none of the previous work has considered the dynamic nature of real-world systems. In this work, we present the first study of adversarial attack on dynamic network link prediction (DNLP). The proposed attack method, namely time-aware gradient attack (TGA), utilizes the gradient information generated by deep dynamic network embedding (DDNE) across different snapshots to rewire a few links, so as to make DDNE fail to predict target links. We implement TGA in two ways: one is based on traversal search, namely TGA-Tra; and the other is simplified with greedy search for efficiency, namely TGA-Gre. We conduct comprehensive experiments which show the outstanding performance of TGA in attacking DNLP algorithms.

Performance troubleshooting is notoriously difficult for distributed microservices-based applications. A typical root-cause diagnosis for performance anomaly by an analyst starts by narrowing down the scope of slow services, investigates into high-level performance metrics or available logs in the slow components, and finally drills down to an actual cause. This process can be long, tedious, and sometimes aimless due to the lack of domain knowledge and the sheer number of possible culprits. This paper introduces a new machine-learning-driven performance analysis system called LongTale that automates the troubleshooting process for latency-related performance anomalies to facilitate the root cause diagnosis and explanation. LongTale builds on existing application-layer tracing in two significant aspects. First, it stitches application-layer traces with corresponding system stack traces, which enables more informative root-cause analysis. Second, it utilizes a novel machine-learning-driven analysis that feeds on the combined data to automatically uncover the most likely contributing factor(s) for given performance slowdown. We demonstrate how LongTale can be utilized in different scenarios, including abnormal long-tail latency explanation and performance interference analysis.

Malware detection is an essential step in building trustworthy computer systems. Signature-based detection detects a sample as malware if the sample data match or contain a pre-stored malware signature. Among all detection methods that malware experts are constantly exploring, signature-based malware detection is indispensable, due to its simplicity, explainability and efficiency. Malware signatures could have various formats, for example, a substring, a subsequence, or a combination rule. A combination rule signature could be viewed as a fixed set of properties, each of which describes some characteristic of an analyzed sample. Although security experts have dedicated many efforts to extract meaningful features from samples, the step of signature generation from the features has been rather ad hoc and time-consuming. This paper focuses on the generation of combination rule signatures. We abstract and formally define the problem of combination rule malware signature generation, followed by a systematic study towards an effective and efficient implementation. Inspired by classic frequent itemsets mining solutions, the proposed AutoCombo approach is greedy but also complete. It generates higher quality signatures first, but is also able to traverse all possible property combinations for a complete generation. Further optimizations and future research potential are also discussed. The proposed approach is currently in use to assist the analysis for millions of files per day in a large security company. Our evaluation results using large-scale production data have also shown its efficacy. With the release of over 10 million real production records as well as our exploratory code, we hope this initial study could draw AI experts' attention and advance the research even further in this field.

A trojan backdoor is a hidden pattern typically implanted in a deep neural network (DNN). It could be activated and thus forces that infected model to behave abnormally when an input sample with a particular trigger is fed to that model. As such, given a DNN and clean input samples, it is challenging to inspect and determine the existence of a trojan backdoor. Recently, researchers design and develop several pioneering solutions to address this problem. They demonstrate that the proposed techniques have great potential in trojan detection. However, we show that none of these existing techniques completely address the problem. On the one hand, they mostly work under an unrealistic assumption of assuming the availability of the contaminated training database. On the other hand, these techniques can neither accurately detect the existence of trojan backdoors, nor restore high-fidelity triggers, especially when infected models are trained with high-dimensional data, and the triggers pertaining to the trojan vary in size, shape, and position. In this work, we propose TABOR, a new trojan detection technique. Conceptually, it formalizes the detection of a trojan backdoor as solving an optimization objective function. Different from the existing technique which also models trojan detection as an optimization problem, TABOR first designs a new objective function that could guide optimization to identify a trojan backdoor more correctly and accurately. Second, TABOR borrows the idea of interpretable AI to further prune the restored triggers. Last, TABOR designs a new anomaly detection method, which could not only facilitate the identification of intentionally injected triggers but also filter out false alarms (i.e., triggers detected from an uninfected model). We train 112 DNNs on five datasets and infect these models with two existing trojan attacks. We evaluate TABOR by using these infected models, and demonstrate that TABOR has much better performance in trigger restoration, trojan detection, and elimination than Neural Cleanse, the state-of-the-art trojan detection technique.

Class distribution skews in imbalanced datasets may lead to models with prediction bias towards majority classes, making fair assessment of classifiers a challenging task. Metrics such as Balanced Accuracy are commonly used to evaluate a classifier's prediction performance under such scenarios. However, these metrics fall short when classes vary in importance. In this paper, we propose a simple and general-purpose evaluation framework for imbalanced data classification that is sensitive to arbitrary skews in class cardinalities and importances. Experiments with several state-of-the-art classifiers tested on real-world datasets from three different domains show the effectiveness of our framework - not only in evaluating and ranking classifiers, but also training them.

While distributed application-layer tracing is widely used for performance diagnosis in microservices, its coarse granularity at the service level limits its applicability towards detecting more fine-grained system level issues. To address this problem, cross-layer stitching of tracing information has been proposed. However, all existing cross-layer stitching approaches either require modification of the kernel or need updates in the application-layer tracing library to propagate stitching information, both of which add further complex modifications to existing tracing tools. This paper introduces Deepstitch, a deep learning based approach to stitch cross-layer tracing information without requiring any changes to existing application layer tracing tools. Deepstitch leverages a global view of a distributed application composed of multiple services and learns the global system call sequences across all services involved. This knowledge is then used to stitch system call sequences with service-level traces obtained from a deployed application. Our proof of concept experiments show that the proposed approach successfully maps application-level interaction into the system call sequences and can identify thread-level interactions.

Outlier detection and novelty detection are two important topics for anomaly detection. Suppose the majority of a dataset are drawn from a certain distribution, outlier detection and novelty detection both aim to detect data samples that do not fit the distribution. Outliers refer to data samples within this dataset, while novelties refer to new samples. In the meantime, backdoor poisoning attacks for machine learning models are achieved through injecting poisoning samples into the training dataset, which could be regarded as "outliers" that are intentionally added by attackers. Differential privacy has been proposed to avoid leaking any individual's information, when aggregated analysis is performed on a given dataset. It is typically achieved by adding random noise, either directly to the input dataset, or to intermediate results of the aggregation mechanism. In this paper, we demonstrate that applying differential privacy can improve the utility of outlier detection and novelty detection, with an extension to detect poisoning samples in backdoor attacks. We first present a theoretical analysis on how differential privacy helps with the detection, and then conduct extensive experiments to validate the effectiveness of differential privacy in improving outlier detection, novelty detection, and backdoor attack detection.

Nowadays an emerging class of applications are based oncollaboration over a shared database among different entities. However, the existing solutions on shared database may require trust on others, have high hardware demand that is unaffordable for individual users, or have relatively low performance. In other words, there is a trilemma among security, compatibility and efficiency. In this paper, we present FalconDB, which enables different parties with limited hardware resources to efficiently and securely collaborate on a database. FalconDB adopts database servers with verification interfaces accessible to clients and stores the digests for query/update authentications on a blockchain. Using blockchain as a consensus platform and a distributed ledger, FalconDB is able to work without any trust on each other. Meanwhile, FalconDB requires only minimal storage cost on each client, and provides anywhere-available, real-time and concurrent access to the database. As a result, FalconDB over-comes the disadvantages of previous solutions, and enables individual users to participate in the collaboration with high efficiency, low storage cost and blockchain-level security guarantees.

While distributed application-layer tracing is widely used for performance diagnosis in microservices, its coarse granularity at the service level limits its applicability towards detecting more fine-grained system level issues. To address this problem, cross-layer stitching of tracing information has been proposed. However, all existing cross-layer stitching approaches either require modification of the kernel or need updates in the application-layer tracing library to propagate stitching information, both of which add further complex modifications to existing tracing tools. This paper introduces Deepstitch, a deep learning based approach to stitch cross-layer tracing information without requiring any changes to existing application layer tracing tools. Deepstitch leverages a global view of a distributed application composed of multiple services and learns the global system call sequences across all services involved. This knowledge is then used to stitch system call sequences with service-level traces obtained from a deployed application. Our proof of concept experiments show that the proposed approach successfully maps application-level interaction into the system call sequences and can identify thread-level interactions.

Community detection, which focuses on cluster vertex interactions, plays a significant role in network analysis. However, it also faces numerous challenges like data missing and adversarial attack. How to further improve the performance and robustness of community detection for real-world networks has raised great concerns. In this paper, a concept of robust enhancement is proposed for community detection, with two algorithms presented: one is named robust enhancement via genetic algorithm (RobustECD-GA), in which the modularity and the number of clusters are used to design a fitness function to solve the resolution limit problem; the other is called robust enhancement via similarity ensemble (RobustECD-SE), integrating multiple information of community structures captured by various vertex similarities, which scales well on large-scale networks. Comprehensive experiments on real-world networks demonstrate, by comparing with two traditional enhancement strategies, that the new methods help six representative community detection algorithms achieve more significant performance improvement. Moreover, experiments on the corresponding adversarial networks indicate that the new methods could also optimize the network structure to a certain extent, achieving stronger robustness against adversarial attack.

Kernel-resident malware remains a significant threat. An effective way to detect such malware is to examine the kernel memory of many similar (virtual) machines, as one might find in an enterprise network or cloud, in search of anomalies: i.e., the relatively rare infected hosts within a large population of healthy hosts. It is challenging, however, to compare the kernel memories of different hosts against each other. Previous work has relied on knowledge of specific kernels-e.g., the locations of important variables and the layouts of key data structures-to cross the "semantic gap" and allow kernels to be compared. As a result, those previous systems work only with the kernels they were built for, and they make assumptions about the malware being searched for. We present a new approach to detecting kernel-resident malware within a "herd" of similar virtual machines. Our approach uses limited knowledge of the kernels under examination-e.g., the location of the page global directory and the processor's instruction set-to concisely fingerprint each kernel. It uses no kernel-specific semantics to compare the fingerprints and find those that represent anomalous hosts. We implement our method in a tool called Fluorescence and demonstrate its ability to identify Linux and Windows hosts infected with real-world, kernel-resident malware. Fluorescence can examine a herd of 200 virtual machines with Linux guests in about an hour.

Differential privacy has been proposed to avoid leaking individual’s information while performing aggregated analysis on a given dataset. It is typically achieved by adding random noise, either directly to the input dataset, or to intermediate results of the aggregation mechanism. The added noise inevitably brings utility downgrade as shown by many previous work, despite its ability to hide individual data. In this paper, we demonstrate that applying differential privacy could surprisingly improve the performance of a type of anomaly detection approach: zero-positive anomaly detection. Specifically, our experimental results indicate that the number of false negatives are significantly reduced, without introducing many false positives.

Anomaly detection is essential towards ensuring system security and reliability. Powered by constantly generated system data, deep learning has been found both effective and flexible to use, with its ability to extract patterns without much domain knowledge. Existing anomaly detection research focuses on a scenario referred to as zero-positive, which means that the detection model is only trained for normal (i.e., negative) data. In a real application scenario, there may be additional manually inspected positive data provided after the system is deployed. We refer to this scenario as lifelong anomaly detection. However, we find that existing approaches are not easy to adopt such new knowledge to improve system performance. In this work, we are the first to explore the lifelong anomaly detection problem, and propose novel approaches to handle corresponding challenges. In particular, we propose a framework called unlearning, which can effectively correct the model when a false negative (or a false positive) is labeled. To this aim, we develop several novel techniques to tackle two challenges referred to as exploding loss and catastrophic forgetting. In addition, we abstract a theoretical framework based on generative models. Under this framework, our unlearning approach can be presented in a generic way to be applied to most zero-positive deep learning-based anomaly detection algorithms to turn them into corresponding lifelong anomaly detection solutions. We evaluate our approach using two state-of-the-art zero-positive deep learning anomaly detection architectures and three real-world tasks. The results show that the proposed approach is able to significantly reduce the number of false positives and false negatives through unlearning.

A trojan backdoor is a hidden pattern typically implanted in a deep neural network. It could be activated and thus forces that infected model behaving abnormally only when an input data sample with a particular trigger present is fed to that model. As such, given a deep neural network model and clean input samples, it is very challenging to inspect and determine the existence of a trojan backdoor. Recently, researchers design and develop several pioneering solutions to address this acute problem. They demonstrate the proposed techniques have a great potential in trojan detection. However, we show that none of these existing techniques completely address the problem. On the one hand, they mostly work under an unrealistic assumption (e.g. assuming availability of the contaminated training database). On the other hand, the proposed techniques cannot accurately detect the existence of trojan backdoors, nor restore high-fidelity trojan backdoor images, especially when the triggers pertaining to the trojan vary in size, shape and position. In this work, we propose TABOR, a new trojan detection technique. Conceptually, it formalizes a trojan detection task as a non-convex optimization problem, and the detection of a trojan backdoor as the task of resolving the optimization through an objective function. Different from the existing technique also modeling trojan detection as an optimization problem, TABOR designs a new objective function--under the guidance of explainable AI techniques as well as heuristics--that could guide optimization to identify a trojan backdoor in a more effective fashion. In addition, TABOR defines a new metric to measure the quality of a trojan backdoor identified. Using an anomaly detection method, we show the new metric could better facilitate TABOR to identify intentionally injected triggers in an infected model and filter out false alarms......

System event logs have been frequently used as a valuable resource in data-driven approaches to enhance system health and stability. A typical procedure in system log analytics is to first parse unstructured logs to structured data, and then apply data mining and machine learning techniques and/or build workflow models from the resulting structured data. Previous work on parsing system event logs focused on offline, batch processing of raw log files. But increasingly, applications demand online monitoring and processing. As a result, a streaming method to parse unstructured logs is needed. We propose an online streaming method Spell, which utilizes a longest common subsequence based approach, to parse system event logs. We show how to dynamically extract log patterns from incoming logs and how to maintain a set of discovered message types in streaming fashion. An enhancement to find more accurate message types is also proposed. We also propose and evaluate a method to automatically discover semantic meanings for parameter fields identified by Spell. We compare Spell against state-of-the-art methods to extract patterns from system event logs on large real data. The results demonstrate that, compared with other log parsing alternatives, Spell shows its superiority in terms of both efficiency and effectiveness.

Federated learning is a recently proposed paradigm that enables multiple clients to collaboratively train a joint model. It allows clients to train models locally, and leverages the parameter server to generate a global model by aggregating the locally submitted gradient updates at each round. Although the incentive model for federated learning has not been fully developed, it is supposed that participants are able to get rewards or the privilege to use the final global model, as a compensation for taking efforts to train the model. Therefore, a client who does not have any local data has the incentive to construct local gradient updates in order to deceive for rewards. In this paper, we are the first to propose the notion of free rider attacks, to explore possible ways that an attacker may construct gradient updates, without any local training data. Furthermore, we explore possible defenses that could detect the proposed attacks, and propose a new high dimensional detection method called STD-DAGMM, which particularly works well for anomaly detection of model parameters. We extend the attacks and defenses to consider more free riders as well as differential privacy, which sheds light on and calls for future research in this field.

Value set analysis (VSA) is one of the most powerful binary analysis tools, which has been broadly adopted in many use cases, ranging from verifying software properties (e.g., variable range analysis) to identifying software vulnerabilities (e.g., buffer overflow detection). Using it to facilitate data flow analysis in the context of postmortem program analysis, it however exhibits an insufficient capability in handling memory alias identification. Technically speaking, this is due to the fact that VSA needs to infer memory reference based on the context of a control flow, but accidental termination of a running program left behind incomplete control flow information, making memory alias analysis clueless. To address this issue, we propose a new technical approach. At the high level, this approach first employs a layer of instruction embedding along with a bi-directional sequence-to-sequence neural network to learn the machine code pattern pertaining to memory region accesses. Then, it utilizes the network to infer the memory region that VSA fails to recognize. Since the memory references to different regions naturally indicate the non-alias relationship, the proposed neural architecture can facilitate the ability of VSA to perform better alias analysis. Different from previous research that utilizes deep learning for other binary analysis tasks, the neural network proposed in this work is fundamentally novel. Instead of simply using off-the-shelf neural networks, we introduce a new neural network architecture which could capture the data dependency between and within instructions. In this work, we implement our deep neural architecture as DEEPVSA, a neural network assisted alias analysis tool. To demonstrate the utility of this tool, we use it to analyze software crashes corresponding to 40 memory corruption vulnerabilities archived in Offensive Security Exploit Database. We show that, DEEPVSA can significantly improve VSA with respect to its capability in analyzing memory alias and thus escalate the ability of security analysts to pinpoint the root cause of software crashes. In addition, we demonstrate that our proposed neural network outperforms state-of-the-art neural architectures broadly adopted in other binary analysis tasks. Last but not least, we show that DEEPVSA exhibits nearly no false positives when performing alias analysis.