526 publications from this institution
Although malicious software (malware) has been around since the early days of computers, the sophistication and innovation of malware has increased over the years. In particular, the latest crop of ransomware has drawn attention to the dangers of malicious software, which can cause harm to private users as well as corporations, public services (hospitals and transportation systems), governments, and security institutions. To protect these institutions and the public from malware attacks, malicious activity must be detected as early as possible, preferably before it conducts its harmful acts. However, it is not always easy to know what to look for—especially when dealing with new and unknown malware that has never been seen. Analyzing a suspicious file by static or dynamic analysis methods can provide relevant and valuable information regarding a file's impact on the hosting system and help determine whether the file is malicious or not, based on the method's predefined rules. While various techniques (e.g., code obfuscation, dynamic code loading, encryption, and packing) can be used by malware writers to evade static analysis (including signature-based anti-virus tools), dynamic analysis is robust to these techniques and can provide greater understanding regarding the analyzed file and consequently can lead to better detection capabilities. Although dynamic analysis is more robust than static analysis, existing dynamic analysis tools and techniques are imperfect, and there is no single tool that can cover all aspects of malware behavior. The most recent comprehensive survey performed in this area was published in 2012. Since that time, the computing environment has changed dramatically with new types of malware (ransomware, cryptominers), new analysis methods (volatile memory forensics, side-channel analysis), new computing environments (cloud computing, IoT devices), new machine-learning algorithms, and more. The goal of this survey is to provide a comprehensive and up-to-date overview of existing methods used to dynamically analyze malware, which includes a description of each method, its strengths and weaknesses, and its resilience against malware evasion techniques. In addition, we include an overview of prominent studies presenting the usage of machine-learning methods to enhance dynamic malware analysis capabilities aimed at detection, classification, and categorization.
No abstract is provided for this article.
Scientific writing is difficult. It is even harder for those for whom English is a second language (ESL learners). Scholars around the world spend a significant amount of time and resources proofreading their work before submitting it for review or publication. In this paper we present a novel machine learning based application for proper word choice task. Proper word choice is a generalization the lexical substitution (LS) and grammatical error correction (GEC) tasks. We demonstrate and evaluate the usefulness of applying bidirectional Long Short Term Memory (LSTM) tagger, for this task. While state-of-the-art grammatical error correction uses error-specific classifiers and machine translation methods, we demonstrate an unsupervised method that is based solely on a high quality text corpus and does not require manually annotated data. We use a bidirectional Recurrent Neural Network (RNN) with LSTM for learning the proper word choice based on a word's sentential context. We demonstrate and evaluate our application on both a domain-specific (scientific), writing task and a general-purpose writing task. We show that our domain-specific and general-purpose models outperform state-of-the-art general context learning. As an additional contribution of this research, we also share our code, pre-trained models, and a new ESL learner test set with the research community.
Sometimes voters are required to reach a joint decision and find an item that best suits the group’s preferences. Voters may wish to state preferences only when necessary, particularly in cases where there are many available options, therefore it is unpractical to assume that all voter preferences are known at all times. In order to elicit voter preferences at a minimal cost, a preference elicitation process is required. We introduce a general approach for reaching a joint decision with minimal elicitation of voter preferences. The approach is probabilistic and uses voting rules to find a necessary winning item which is presented to the group as their best option. We propose computing a voter-item probability distribution and developing methods based on this distribution that can then determine which voter-item pair to query. Computing the optimal minimal set of voter-item queries is computationally intractable; therefore we propose novel heuristic algorithms, named DIG and ES, which proceed iteratively until the identification of a winning item. The probabilistic voting distribution is updated as more information is revealed. Experiments on simulated data examine the benefits of each of the algorithms under different settings. Experiments with the real-world Netflix data show that the proposed algorithms reduce the required number of ratings for identifying the winning item by more than 50%.
Topic: 30. Infections in hematology (incl. supportive care/therapy) Background: Between 2020 and 2022, COVID-19 affected more than 70% of the Israeli population. It is now well known that haemato-oncologic patients are more susceptible to severe infections, including persistent COVID, and have a lower immune-response to COVID-19 vaccines than the general population. However, there is limited research on the outcomes of COVID-19 infection among patients with primary cutaneous lymphomas, such as Mycosis Fungoides and Sezary Syndrome (MF/SS). Aims: We aimed to assess the clinical outcomes of COVID-19 among patients with MF/SS during the period of 2020-2022. Methods: The data was retrieved from anonymized electronic medical records of Maccabi Healthcare Services, the second largest healthcare organization in Israel, after receiving approval from the institute’s ethical committee. Patients diagnosed with MF/SS by dermatologists, hematologists, or plastic surgeons and coded according to the ICD-9 system were included in the study. COVID-19 PCR test results, socio-demographic and clinical data were extracted and analyzed. Correlations tests and multivariate analyses were used to evaluate the association between COVID-19 infection, different COVID-19 variants, treatment, vaccination status, and clinical outcomes. Results: A total of 1472 patients diagnosed with MF/SS between 2020 and 2022 were included in the study. Among them, 768 (52%) had at least one confirmed SARS-CoV-2 infection by a PCR test, which was more prevalent among younger patients (p<0.001) and patients with comorbidities such as hypertension (p=0.009), overweight (p=0.001), kidney failure (p=0.046), osteoporosis (p=0.003) or dementia (p=0.046). The hospitalization rate was 2.9%, and the 30-day mortality rate from any cause after COVID-19 infection was 1.2% during the study period. Delta had the highest hospitalization rate (7.7%), among the three main COVID-19 variants, followed by Alpha (2.4%) and then Omicron (2.0%). During the Delta and Omicron surges, the hospitalization rate was lower among fully vaccinated patients, compared to non-vaccinated or non-fully vaccinated (p=0.032). However, it was higher among patients older than 65 (p<0.001) and patients with SS (vs. MF) (p<0.001) or COPD (p=0.024) diagnosis. Since January 2, 2022, the first day Nirmatrelvir was available for Maccabi patients, nirmatrelvir + ritonavir was administrated to 76 (11%) patients. There was a tendency for decreased hospitalization among patients treated within five days of infection, with a 79% risk reduction, although it was not statistically significant (p=0.164). Summary/Conclusion: It seems that patients with MF/SS do not necessarily have worse COVID-19 outcomes compared to the general population. However, further studies are needed to evaluate the effects of the different MF/SS treatments on COVID-19 severity and antibody formation to vaccination.Keywords: Mycosis fungoides, COVID-19
Anomaly detection is a well-known task that involves the identification of abnormal events that occur relatively infrequently. Methods for improving anomaly detection performance have been widely studied. However, no studies utilizing test-time augmentation (TTA) for anomaly detection in tabular data have been performed. TTA involves aggregating the predictions of several synthetic versions of a given test sample; TTA produces different points of view for a specific test instance and might decrease its prediction bias. We propose the Test-Time Augmentation for anomaly Detection (TTAD) technique, a TTA-based method aimed at improving anomaly detection performance. TTAD augments a test instance based on its nearest neighbors; various methods, including the k-Means centroid and SMOTE methods, are used to produce the augmentations. Our technique utilizes a Siamese network to learn an advanced distance metric when retrieving a test instance’s neighbors. Our experiments show that using our TTA technique significantly improves the performance of anomaly detection algorithms, as evidenced by the higher AUC results achieved on all datasets evaluated. Specifically, we observed average improvements of 0.037 AUC (3.7%) using Autoencoder, 0.016 AUC (1.6%) using OC-SVM, and 0.023 AUC (2.3%) using LOF.
Can a machine-learning-based model trained in clinical and biological variables support the prediction of the presence or absence of sperm in testicular biopsy in non-obstructive azoospermia (NOA) patients?Our machine-learning model was able to accurately predict (AUC of 0.8) the presence or absence of spermatozoa in patients with NOA.Patients with NOA can conceive with their own biological gametes using ICSI in combination with successful testicular sperm extraction (TESE). Testicular sperm retrieval is successful in up to 50% of men with NOA. However, to the best of our knowledge, there is no existing model that can accurately predict the success of sperm retrieval in TESE. Moreover, machine-learning has never been used for this purpose.A retrospective cohort study of 119 patients who underwent TESE in a single IVF unit between 1995 and 2017 was conducted. All patients with NOA who underwent TESE during their fertility treatments were included. The development of gradient-boosted trees (GBTs) aimed to predict the presence or absence of spermatozoa in patients with NOA. The accuracy of these GBTs was then compared to a similar multivariate logistic regression model (MvLRM).We employed univariate and multivariate binary logistic regression models to predict the probability of successful TESE using a dataset from a retrospective cohort. In addition, we examined various ensemble machine-learning models (GBT and random forest) and evaluated their predictive performance using the leave-one-out cross-validation procedure. A cutoff value for successful/unsuccessful TESE was calculated with receiver operating characteristic (ROC) curve analysis.ROC analysis resulted in an AUC of 0.807 ± 0.032 (95% CI 0.743-0.871) for the proposed GBTs and 0.75 ± 0.052 (95% CI 0.65-0.85) for the MvLRM for the prediction of presence or absence of spermatozoa in patients with NOA. The GBT approach and the MvLRM yielded a sensitivity of 91% vs. 97%, respectively, but the GBT approach has a specificity of 51% compared with 25% for the MvLRM. A total of 78 (65.3%) men with NOA experienced successful TESE. FSH, LH, testosterone, semen volume, age, BMI, ethnicity and testicular size on clinical evaluation were included in these models.This study is a retrospective cohort study, with all the associated inherent biases of such studies. This model was used only for TESE, since micro-TESE is not performed at our center.Machine-learning models may lay the foundation for a decision support system for clinicians together with their NOA patients concerning TESE. The findings of this study should be confirmed with further larger and prospective studies.The study was funded by the Division of Obstetrics and Gynecology, Soroka University Medical Center, there are no potential conflicts of interest for all authors.
Providing adaptive help during interaction with the system can be used to assist users in accomplishing their tasks. We propose providing guidance by highlighting the steps required for performing a task that the user intends to complete according to the prediction of a system. We present a study aimed at examining whether highlighting intended user steps in menus and toolbars as a means of assisting users in performing tasks is useful in terms of user response and performance. We also examined the effects of different accuracy levels of help and the control format on user response and performance. An experiment was conducted in which 64 participants performed tasks using menus and toolbars of a simulated email application. Participants were offered a highlighted guidance of the required steps in varying levels of accuracy (100%, 80%, 60% and no guidance). Our results support the benefits of highlighted help both in user performance times and in user satisfaction from receiving such assistance. Users found the assistance necessary and helpful and by the same token not unduly intrusive. Additionally, users felt that such assistance generally helped in reducing performance time on tasks. We did not find a significant difference when users receiving help at 80% accuracy was compared to those receiving help at 100% accuracy; however, such a difference does appear for those receiving 60% accuracy. In such cases, we found that the user's satisfaction level, perceived usefulness and trust in the system decreased while their notion of perceived intrusiveness increased. We conclude that assisting users by highlighting the required steps is useful so long as the minimal accuracy level of help is higher than 60%. Our study has implications on the implementation of highlighting next steps as a means of adaptive help and on integrating probability-based algorithms such as intention prediction to adaptive assistance systems. Keywords: adaptive assistancehighlighted helpaccuracy levelmenus and toolbars
No abstract is provided for this article.
Matrix factorization methods such as the singular value decomposition technique have become very popular in the area of recommender systems. Given a rating matrix as input, these techniques output two matrixes with lower dimensional space that represent the user and item features. The relevance of item i to user u is revealed by the score of the dot product between u vector of features and i vector of features. High scores indicate greater relevance. In order to deliver the best recommendations for a given user based on these latent features, one must obtain the list of scores of all the items for the given user and sort the resulting list. When the size of the catalogue is large, this phase consumes a large amount of computational time and cannot be done online. Another drawback with this approach is that once such a list is computed for a given user, it remains finite and it is impossible to incorporate within it new activities of the user. Hence, the use of such techniques is limited online. In this paper we propose an ensemble method for building a forest of trees offline, where each leaf in each tree is holding a unique set of item vectors. Once a user is engaged with the system, its vector is classified to one leaf in each one of the trees in the forest for conducting a dot product with the corresponding items. By using this method we compute online only a small number of dot products for a given user vector allowing us to quickly retrieve dynamic recommendations from the SVD, thereby presenting an alternative to the existing method which computes and caches all of the dot products among the items and users. The method maps the items to the leaves of multiple compact trees offline, each tree is a weak recommendation model, creating a forest of decision trees algorithm in which users that are assigned to these leaves online are likely to produce high dot product scores with the items that are already in the leaves. We demonstrate the effectiveness of the suggested ensemble method by applying it to three public datasets and comparing it to a state-of-the-art algorithm aimed at solving the problem.
This study introduces BagStacking, an innovative ensemble learning framework designed to enhance the detection of freezing of gait (FOG) in Parkinson’s disease (PD) using accelerometer data. By synergistically combining bagging’s variance reduction with stacking’s sophisticated blending mechanisms, BagStacking achieves superior predictive performance. Evaluated on a comprehensive PD dataset provided by the Michael J. Fox Foundation, BagStacking attained a mean average precision (MAP) of 0.306, surpassing standalone LightGBM and traditional stacking methods. Furthermore, BagStacking demonstrated superior area under the curve (AUC) metrics across key FOG event classes. Specifically, it achieved AUCs of 0.88 for start hesitation, 0.90 for turning, and 0.84 for walking events, outperforming multistrategy ensemble, regular stacking, and LightGBM baselines. Additionally, BagStacking exhibited reduced runtime compared to other ensemble approaches, making it suitable for real-time clinical monitoring. These results underscore BagStacking’s effectiveness in addressing the variability inherent in FOG detection, thereby contributing to improved patient care in PD.
With the growing number of machine learning (ML) algorithms, the selection of the top-performing algorithms for a given dataset, task, and evaluation measure is known to be a challenging task. The human expertise required for this task has fueled the demand for automatic solutions. Meta-learning is a popular approach for automatic algorithm selection based on dataset characterization. Existing meta-learning methods often represent the datasets using predefined features and thus cannot be generalized for various ML tasks, or alternatively, learn their representations in a supervised fashion, and thus cannot address unsupervised tasks. In this study, we first propose a novel learning-based task-agnostic method for dataset representation. Second, we present TRIO, a meta-learning approach based on the proposed dataset representation, which is capable of accurately recommending top-performing algorithms for unseen datasets. TRIO first learns graphical representations from the datasets and then utilizes a graph convolutional neural network technique to extract their latent representations. An extensive evaluation on 337 datasets and 195 ML algorithms demonstrates the effectiveness of our approach over state-of-the-art methods for algorithm selection for both supervised (classification and regression) and unsupervised (clustering) tasks.
