A refined decompiler to generate C code with high readability

Recent advances in microarray technology have opened new ways for functional annotation of previously uncharacterised genes on a genomic scale. This has been demonstrated by unsupervised clustering of co-expressed genes and, more importantly, by supervised learning algorithms. Using prior knowledge, these algorithms can assign functional annotations based on more complex expression signatures found in existing functional classes. Previously, support vector machines (SVMs) and other machine-learning methods have been applied to a limited number of functional classes for this purpose. Here we present, for the first time, the comprehensive application of supervised neural networks (SNNs) for functional annotation. Our study is novel in that we report systematic results for ~100 classes in the Munich Information Center for Protein Sequences (MIPS) functional catalog. We found that only ~10% of these are learnable (based on the rate of false negatives). A closer analysis reveals that false positives (and negatives) in a machine-learning context are not necessarily "false" in a biological sense. We show that the high degree of interconnections among functional classes confounds the signatures that ought to be learned for a unique class. We term this the "Borges effect" and introduce two new numerical indices for its quantification. Our analysis indicates that classification systems with a lower Borges effect are better suitable for machine learning. Furthermore, we introduce a learning procedure for combining false positives with the original class. We show that in a few iterations this process converges to a gene set that is learnable with considerably low rates of false positives and negatives and contains genes that are biologically related to the original class, allowing for a coarse reconstruction of the interactions between associated biological pathways. We exemplify this methodology using the well-studied tricarboxylic acid cycle.

The extensive spread of harmful content, including hate speech, harassment, violent and adult material across online platforms and media channels poses significant challenges and has raised considerable concern within various societal groups. Government bodies, educators, and parents frequently find themselves in disagreement with media providers over the best methods to regulate and restrict such content. Automated content moderation technologies have become critical tools in addressing these challenges, particularly through natural language processing (NLP) techniques that can automatically detect and filter sensitive textual content, such as offensive language, violence, and adult materials, enabling platforms to enforce moderation policies on a large scale. Despite their widespread use, current moderation technologies face challenges related to detection accuracy, often resulting in substantial false positives and false negatives. To enhance content moderation systems, more advanced algorithms capable of accurately interpreting textual context are necessary. In this study, we assess current large language model (LLM)-based moderation solutions, specifically OpenAI moderation model and Llama-Guard-3, examining their effectiveness in detecting sensitive content. Additionally, we investigate the capabilities of contemporary LLMs, including OpenAI generative pre-trained transformer (GPT), Google Gemini, Meta Llama, Anthropic Claude, and small language models (SLMs) such as Google Gemma, in recognizing inappropriate content from diverse media sources. We also studied the performance of these models under adversarial attacks such as input perturbation and prompt injection. Our evaluation and comparative analysis utilize various textual datasets, such as X (Twitter) posts, Amazon product reviews, and news articles. The findings indicate that LLM-based approaches significantly demonstrate high accuracy and low rates of false positives and negatives. They also indicate the robustness of the models under various adversarial attacks. These results underscore the considerable potential for integrating advanced LLMs into websites and social media platforms, thereby enhancing content regulation and moderation effectiveness.

Measuring the internal quality of source code is one of the traditional goals of making software development into an engineering discipline. Cyclomatic complexity (CC) is an often used source code quality metric, next to source lines of code (SLOC). However, the use of the CC metric is challenged by the repeated claim that CC is redundant with respect to SLOC because of strong linear correlation.We conducted an extensive literature study of the CC/SLOC correlation results. Next, we tested correlation on large Java (17.6 M methods) and C (6.3 M functions) corpora. Our results show that linear correlation between SLOC and CC is only moderate as a result of increasingly high variance. We further observe that aggregating CC and SLOC as well as performing a power transform improves the correlation.Our conclusion is that the observed linear correlation between CC and SLOC of Java methods or C functions is not strong enough to conclude that CC is redundant with SLOC. This conclusion contradicts earlier claims from literature but concurs with the widely accepted practice of measuring of CC next to SLOC. Copyright © 2015 John Wiley & Sons, Ltd.

Background: High plasma concentrations of lipoprotein (a) [Lp(a)] are associated with an increased cardiovascular risk. Current guidelines recommend measurement of only a single Lp(a) in an individual’s lifetime under specific circumstances to improve cardiovascular risk prediction. Accordingly, the question raised is the number of false positives and negatives missed through only a single measurement. Methods: All Lp(a) measurements between 2004 and March 2021 were retrieved from the laboratory database of the Erasme hospital. Only patients with repeated measurement were included. The first and subsequent Lp(a) measurement were compared. Two different cohorts were studied as a result of a change in Lp(a) determination methodology (n = 2049 and n = 309, respectively). The effects of a third Lp(a) measurement were assessed through binary analyses (n = 678). The 180 mg/dl (430 nmol/L) threshold recommended in the ESC guidelines was assessed first. Analysis was repeated for 100, 70 and 50 mg/dl thresholds of raised Lp(a) levels. Results: A low rate of false negatives (0.8%–1%) and false positives (0.6–0.3%) were revealed with two Lp(a) measurements. There was no difference in regards to the divergent Lp(a) thresholds nor the measurement of Lp(a) on two or three occasions. Conclusion: The present study showed Lp(a) determination to be reproducible. A single measurement is sufficient to assess if a patient exceeds various cut-off values of elevated Lp(a) levels.

Peer-to-Peer system has achieved great success with millions of end users in the past several years. P2P traffic has occupied about 60–80 % of the total traffic volumes, which greatly consumes network bandwidth and causes congestions. To achieve the goal of efficacious P2P system management in the monitored network, in this paper we develop a framework named CUFTI (Core Users Finding and Traffic Identification). The core users are referred as long-lived peers, and we focus on life-time characteristics of coexisting peers within each snapshot of the overlay. Based on the analysis results of user’s behaviour in PPlive system, we develop an accurate model to forecast the peer’s residual life-time and identify the long-lived peers. Furthermore, we develop a flow identification model for P2P traffic management of those core users. Based on the analysis results of actual traffic traces, we find the P2P traffic flows are composed of data and control packets. Most of the control packets appear at the beginning and end of each flow to establish and close the communication between peers. We employ the direction and payload length of the control packets at the beginning of the flow as features to perform flow identification. Experimental results based on traces collected from the Northwest Region Center of CERNET (China Education and Research Network) show that the newly developed methods outperforms other existing methods with lower false negative rate (FNR) and false positive rate (FPR).

Tandem mass spectrometry (MS/MS) has been widely used in proteomics studies. Multiple algorithms have been developed for assessing matches between MS/MS spectra and peptide sequences in databases. However, it is still a challenge to reduce false negative rates without compromising the high confidence of peptide identification. In this study, we developed the score, Oscore, by logistic regression using SEQUEST and AMASS variables to identify fully tryptic peptides. Since these variables showed complicated association with each other, combining them together rather than applying them to a threshold model improved the classification of correct and incorrect peptide identifications. Oscore achieved both a lower false negative rate and a lower false positive rate than PeptideProphet on datasets from 18 known protein mixtures and several proteome-scale samples of different complexity, database size and separation methods. By a three-way comparison among Oscore, PeptideProphet and another logistic regression model which made use of PeptideProphet's variables, the main contributor for the improvement made by Oscore is discussed.

In a Delay-Tolerant Network (DTN), data originating from a source node may be delivered to the destination node, despite the non-existence of end-to-end connectivity between them at all times. In an adversarial environment such as a battlefield, DTN nodes could be compromised to launch Denial-of-Service (DoS) attacks by generating excess data, to cause an overflow of the limited resources of the legitimate nodes, hence decreasing the network throughput. A node may also display selfish behavior by generating more data than allowed, to increase its throughput and to decrease the latency of its data packets. In this paper, we term such a DoS attack and selfish data generation behavior, a resource-misuse attack. We study two types of resource-misuse attacks, breadth attacks and depth attacks. Accordingly, we propose different schemes to detect these attacks. Trace-driven simulations using both a synthetic and a real-world trace show that our detection schemes have low average detection latency and additionally, probabilistic detection of the depth attack has low false positive and false negative rates.

Cybercriminals exploit malicious URLs as a distribution channel to spread harmful software across the internet. They take advantage of vulnerabilities in browsers to install malicious software with the aim of gaining remote access to the victims' computers. Typically, this malicious software aims to gain access to networks, steal sensitive information, and silently monitor targeted computer systems. In this research, a data mining approach known as Classification Based on Association (CBA) is employed to detect malicious URLs using both the URL itself and the features of the presented web pages. The CBA algorithm utilizes a training dataset of URLs as historical data to discover association rules that can be used to create an accurate classifier. By detecting dangerous URLs and malicious software, this contribution can assist organizations and individual users in enhancing the security of their computer systems and networks, thereby protecting sensitive data and reducing the risk of security incidents. The experimental results demonstrate that CBA achieves performance on par with tested classification algorithms, achieving an accuracy of 99% and low rates of false positives and false negatives. Future research could expand its focus to detect malicious URLs and software on mobile devices and embedded systems, as they have become significant targets for cybercriminals.

RAD-OAEnsemble: Residual attention DenseNet with radiograph-EHR fusion and dual-stream deep feature embedding for knee osteoarthritis risk prediction

A botnet, a set of compromised machines controlled distantly by an attacker, is the basis of numerous security threats around the world. Command and Control servers are the backbones of botnet communications, where the bots and botmasters send report and attack orders to each other. Botnets are also categorized according to their C&C protocols. A Domain Name System method known as Fast-Flux Service Network (FFSN) — a special type of botnet — has been engaged by bot herders to cover malicious botnet activities and increase the lifetime of malicious servers by quickly changing the IP addresses of the domain name over time. Although several methods have been suggested for detecting FFSNs, they have low detection accuracy especially with zero-day domain. In this research, we propose a new system called Fast Flux Killer System (FFKS) that has the ability to detect FF-Domains in online mode with an implementation constructed on Adaptive Dynamic evolving Spiking Neural Network (ADeSNN). The proposed system proved its ability to detect FF domains in online mode with high detection accuracy (98.77%) compare with other algorithms, with low false positive and negative rates respectively. It is also proved a high level of performance. Additionally, the proposed adaptation of the algorithm enhanced and helped in the parameters customization process.

In order to find the attack in real time, an intrusion prediction method based on intelligent immune threshold matching algorithm was presented. Using a dynamic load-balancing algorithm, network data packet was distributed to a set of predictors by the balancer; it could avoid packet loss and false negatives in high-performance network with handling heavy traffic loads in real-time. In addition, adopting the dynamic threshold value, which was generated from variable network speed, the mature antibody could better match the antigen of the database, and consequently the accuracy of prediction was increased. Experiment shows this intrusion prediction method has relatively low false positive rate and false negative rate, so it effectively resolves the shortage of intrusion detection.KeywordsIntrusion predictionfalse alarm ratefalse negative rateintelligent immune threshold matching algorithm

: This research aims at the enhancement of computer defenses by making them invulnerable to new, mutating and obfuscated malware. It offers a semantic approach to system behavior analysis, centered on the concept of functionality. Functionality is the highest level of the behavior semantics, it is defined by the specific goal of computer operations, not by its software realization. This allows for identifying some classes of malware achieving the same specific malicious operations. Colored Petri nets are proposed as a basis for behavioral signatures representing particular functionalities, both legitimate and malicious. Special techniques are proposed to address three interrelated aspects: signature expressiveness, behavioral obfuscation and run-time signature matching efficiency. A signature based approach for detecting malicious functionalities in the system call domain is developed. It has been implemented in a prototype software and tested. It is superior to existing behavior based techniques in addressing behavioral obfuscations and multiple functionality realizations. The experiments results indicate low rate of false positives and negatives, and low execution overhead. Such results suggest that detecting malicious functionality presents a sufficiently dependable and efficient method for distinguishing malware from benign software.

To compare quality of harvested tissue, false-negative rate, and complication rate of large-core needle breast biopsy (LCNBB) and directional, vacuum-assisted breast biopsy (DVABB) in non-palpable breast lesions. From 1994 to 1999, in 580 non-palpable breast lesions a stereotactically-guided breast biopsy was performed. 14-G LCNBB was used in 168 lesions (29 %). DVABB was used in 412 lesions (71 %; 11-G DVABB: 134 lesions 32.5 %, 14-G DVABB: 278 lesions 67.5 %). Following biopsy, patients underwent either surgical excision (n = 533; 93.8 %) or mammographical follow-up (n = 36; 6.2 %). Histological results of LCNBB, DVABB, surgical breast biopsy and follow-up results were compared and scored for their tissue quality on a three-point scale (1 = disagreement between biopsy and surgery; 2 = partial agreement; 3 = complete agreement). In addition, we determined the false negative and complication rate for both systems. Histological examination after surgery and follow-up proved 262 (45.2 %) to be benign, 15 (2.6 %) to be high-risk lesions and 303 (52.5 %) to be malignant. In the tissue quality there was no significant difference between 14-G LCNBB (score = 2.94), 11-G DVABB (score = 2.92) and 14-G DVABB (score = 2.91) (p > 0.05). Particularly, in calcifications 11-G DVABB scored better (score = 2.92) than 14-G DVABB (score = 2.88) (p > 0.05). 14-G LCNBB had a lower false negative rate (1.8 %) than 11-G DVABB (3 %) and 14-G DVABB (3.2 %) (p > 0.05). There was no difference in the complication rate between the different needle types. Our results indicate that both LCNBB and DVABB are reliable and safe techniques in the diagnosis of non-palpable breast lesions. However, the use of 14-G LCNBB seems to be advantageous in masses, the use of 11-G DVABB seems to be advantageous in asymmetric densities and calcifications.

BackgroundA composite biological structure, such as an insect head or abdomen, contains many internal structures with distinct functions. Composite structures are often used in RNA-seq studies, though it is unclear how expression of the same gene in different tissues and structures within the same structure affects the measurement (or even utility) of the resulting patterns of gene expression. Here we determine how complex composite tissue structure affects measures of gene expression using RNA-seq.ResultsWe focus on two structures in the honey bee (the sting gland and digestive tract) both contained within one larger structure, the whole abdomen. For each of the three structures, we used RNA-seq to identify differentially expressed genes between two developmental stages, nurse bees and foragers. Based on RNA-seq for each structure-specific extraction, we found that RNA-seq with composite structures leads to many false negatives (genes strongly differentially expressed in particular structures which are not found to be differentially expressed within the composite structure). We also found a significant number of genes with one pattern of differential expression in the tissue-specific extraction, and the opposite in the composite extraction, suggesting multiple signals from such genes within the composite structure. We found these patterns for different classes of genes including transcription factors.ConclusionsMany RNA-seq studies currently use composite extractions, and even whole insect extractions, when tissue and structure specific extractions are possible. This is due to the logistical difficultly of micro-dissection and unawareness of the potential errors associated with composite extractions. The present study suggests that RNA-seq studies of composite structures are prone to false negatives and difficult to interpret positive signals for genes with variable patterns of local expression. In general, our results suggest that RNA-seq on large composite structures should be avoided unless it is possible to demonstrate that the effects shown here do not exist for the genes of interest.

In early breast cancer the status of the axillary nodes has been shown to be one of the primary prognostic indicators. Biopsy of the sentinel node, or first draining lymph node, of a tumour has been investigated as an alternative to axillary dissection in early breast cancer. A series of sentinel node biopsies in 103 patients is reported here. Both pre-operative lymphoscintigraphy and intra-operative blue dye were used to map the sentinel nodes. Mapping was successful in 87 (84.4%) cases and sentinel nodes were retrieved in 94.2% of these patients. Where lymphoscintigraphic mapping was unsuccessful, sentinel nodes were found in 37.5%. When sentinel nodes were retrieved, correlation of the sentinel node status with the axillary nodes was accurate in 97.5%. There were two false negatives, both in large tumours. The sentinel node status was an accurate predictor of axillary status in 95.7% of the node positive patients. If only the 86 patients with invasive carcinoma and four or more axillary nodes removed at surgery are considered, the sentinel node was accurate in assessing the axillary status in 97.7% of the total patient group (2.3% false negative rate), 97.2% of those in whom sentinel nodes were successfully retrieved (2.8% false negative rate) and 94.9% of the patients with positive axillary nodes (5.1% false negative rate). Sentinel node biopsy is a valid technique providing an accurate reflection of the axillary node status and having a low false negative rate.