Rethinking Embedded Blocks for Machine Learning Applications

The underlying goal of FPGA architecture research is to devise flexible substrates that implement a wide variety of circuits efficiently. Contemporary FPGA architectures have been optimized to support networking, signal processing, and image processing applications through high-precision digital signal processing (DSP) blocks. The recent emergence of machine learning has created a new set of demands characterized by: (1) higher computational density and (2) low precision arithmetic requirements. With the goal of exploring this new design space in a methodical manner, we first propose a problem formulation involving computing nested loops over multiply-accumulate (MAC) operations, which covers many basic linear algebra primitives and standard deep neural network (DNN) kernels. A quantitative methodology for deriving efficient coarse-grained compute block architectures from benchmarks is then proposed together with a family of new embedded blocks, called MLBlocks. An MLBlock instance includes several multiply-accumulate units connected via a flexible routing, where each configuration performs a few parallel dot-products in a systolic array fashion. This architecture is parameterized with support for different data movements, reuse, and precisions, utilizing a columnar arrangement that is compatible with existing FPGA architectures. On synthetic benchmarks, we demonstrate that for 8-bit arithmetic, MLBlocks offer 6× improved performance over the commercial Xilinx DSP48E2 architecture with smaller area and delay; and for time-multiplexed 16-bit arithmetic, achieves 2× higher performance per area with the same area and frequency. All source codes and data, along with documents to reproduce all the results in this article, are available at http://github.com/raminrasoulinezhad/MLBlocks .

A Unified Survey of Treatment Effect Heterogeneity Modelling and Uplift Modelling

A central question in many fields of scientific research is to determine how an outcome is affected by an action, i.e., to estimate the causal effect or treatment effect of an action. In recent years, in areas such as personalised healthcare, sociology, and online marketing, a need has emerged to estimate heterogeneous treatment effects with respect to individuals of different characteristics. To meet this need, two major approaches have been taken: treatment effect heterogeneity modelling and uplifting modelling. Researchers and practitioners in different communities have developed algorithms based on these approaches to estimate the heterogeneous treatment effects. In this article, we present a unified view of these two seemingly disconnected yet closely related approaches under the potential outcome framework. We provide a structured survey of existing methods following either of the two approaches, emphasising their inherent connections and using unified notation to facilitate comparisons. We also review the main applications of the surveyed methods in personalised marketing, personalised medicine, and sociology. Finally, we summarise and discuss the available software packages and source codes in terms of their coverage of different methods and applicability to different datasets, and we provide general guidelines for method selection.

A Unified Perspective for Disinformation Detection and Truth Discovery in Social Sensing: A Survey

With the proliferation of social sensing, large amounts of observation are contributed by people or devices. However, these observations contain disinformation. Disinformation can propagate across online social networks at a relatively low cost, but result in a series of major problems in our society. In this survey, we provide a comprehensive overview of disinformation and truth discovery in social sensing under a unified perspective, including basic concepts and the taxonomy of existing methodologies. Furthermore, we summarize the mechanism of disinformation from four different perspectives (i.e., text only, text with image/multi-modal, text with propagation, and fusion models). In addition, we review existing solutions based on these requirements and compare their pros and cons and give a sort of guide to usage based on a detailed lesson learned. To facilitate future studies in this field, we summarize related publicly accessible real-world data sets and open source codes. Last but the most important, we emphasize potential future research topics and challenges in this domain through a deep analysis of most recent methods.

P1OVD: Patch-Based 1-Day Out-of-Bounds Vulnerabilities Detection Tool for Downstream Binaries

In the past decades, due to the popularity of cloning open-source software, 1-day vulnerabilities are prevalent among cyber-physical devices. Detection tools for 1-day vulnerabilities effectively protect users who fail to adopt 1-day vulnerability patches in time. However, manufacturers can non-standardly build the binaries from customized source codes to multiple architectures. The code variants in the downstream binaries decrease the accuracy of 1-day vulnerability detections, especially when signatures of out-of-bounds vulnerabilities contain incomplete information of vulnerabilities and patches. Motivated by the above observations, in this paper, we propose P1OVD, an effective patch-based 1-day out-of-bounds vulnerability detection tool for downstream binaries. P1OVD first generates signatures containing patch information and vulnerability root cause information. Then, P1OVD uses an accurate and robust matching algorithm to scan target binaries. We have evaluated P1OVD on 104 different versions of 30 out-of-bounds vulnerable functions and 620 target binaries in six different compilation environments. The results show that P1OVD achieved an accuracy of 83.06%. Compared to the widely used patch-level vulnerability detection tool ReDeBug, P1OVD ignores 4.07 unnecessary lines on average. The experiments on the x86_64 platform and the O0 optimization show that P1OVD increases the accuracy of the state-of-the-art tool, BinXray, by 8.74%. Besides, it can analyze a single binary in 4 s after a 20-s offline signature extraction on average.

Numerical Investigation into the Influence of Grain Orientation Distribution on the Local and Global Elastic-Plastic Behaviour of Polycrystalline Nickel-Based Superalloy INC-738 LC

Polycrystalline nickel-based superalloys tend to have large grains within component areas where high loads are dominant during operation. Due to these large grains, caused by the manufacturing and cooling process, the orientation of each grain becomes highly important, since it influences the elastic and plastic behaviour of the material. With the usage of the open source codes NEPER and FEPX, polycrystalline models of Inconel 738 LC were generated and their elastic and crystal plasticity behaviour simulated in dependence of different orientation distributions under uniaxial loading. Orientation distributions close to the [100] direction showed the lowest Young’s moduli as well as the highest elastic strains before yielding, as expected. Orientations close to the [5¯89] direction, showed the lowest elastic strains and therefore first plastic deformation under strain loading due to the highest shear stress in the slip systems caused by the interaction of Young’s modulus and the Schmid factor.

RNA-combine: a toolkit for comprehensive analyses on transcriptome data from different sequencing platforms

Background Understanding the transcriptome has become an essential step towards the full interpretation of the biological function of a cell, a tissue or even an organ. Many tools are available for either processing, analysing transcriptome data, or visualizing analysis results. However, most existing tools are limited to data from a single sequencing platform and only several of them could handle more than one analysis module, which are far from enough to meet the requirements of users, especially those without advanced programming skills. Hence, we still lack an open-source toolkit that enables both bioinformatician and non-bioinformatician users to process and analyze the large transcriptome data from different sequencing platforms and visualize the results. Results We present a Linux-based toolkit, RNA-combine, to automatically perform the quality assessment, downstream analysis of the transcriptome data generated from different sequencing platforms, including bulk RNA-seq (Illumina platform), single cell RNA-seq (10x Genomics) and Iso-Seq (PacBio) and visualization of the results. Besides, this toolkit is implemented with at least 10 analysis modules more than other toolkits examined in this study. Source codes of RNA-combine are available on GitHub: https://github.com/dongxuemin666/RNA-combine. Conclusion Our results suggest that RNA-combine is a reliable tool for transcriptome data processing and result interpretation for both bioinformaticians and non-bioinformaticians.

FinTech Companies: A Bibliometric Analysis

The financial-technology industry has recently attracted the attention of many sectors. The financial-technology industry designs new and unusual technological financial services in many areas. It combines technology with finance and provides an alternative to the traditional financial system. In the scope of this study, 636 publications were obtained from Scopus. Various tools, such as Microsoft Excel for frequency analysis, and VOSviewer for data visualization, were used. The open-source codes used for bibliometric analysis through the R Studio program were developed by the authors and used for citation-metrics analysis. The main aim of this study was to find out the most influential studies and authors and to reveal the distributions and impacts of publications in the FinTech area between 2015 and 2021 from the Scopus database. The results indicate that the most influential journal is Sustainability Switzerland, and the most cited author is Gomber et al. Additionally, Rabbani has the most publications, while China has emerged as the most productive country. On the other hand, this study found that FinTech research clustered in four areas. These areas are computer science, business management, economics, and social sciences. This FinTech study examines financial services, financial access, and financial technology, where FinTech is at the center. It also focuses on cryptocurrency, bitcoin, and smart contracts where the blockchain is at the center. The results reveal a systematic map of existing studies. Further, the study plays a guiding role in future research.

An Automated Behaviour-Based Clustering of IoT Botnets

The leaked IoT botnet source-codes have facilitated the proliferation of different IoT botnet variants, some of which are equipped with new capabilities and may be difficult to detect. Despite the availability of solutions for automated analysis of IoT botnet samples, the identification of new variants is still very challenging because the analysis results must be manually interpreted by malware analysts. To overcome this challenge, we propose an approach for automated behaviour-based clustering of IoT botnet samples, aimed to enable automatic identification of IoT botnet variants equipped with new capabilities. In the proposed approach, the behaviour of the IoT botnet samples is captured using a sandbox and represented as behaviour profiles describing the actions performed by the samples. The behaviour profiles are vectorised using TF-IDF and clustered using the DBSCAN algorithm. The proposed approach was evaluated using a collection of samples captured from IoT botnets propagating on the Internet. The evaluation shows that the proposed approach enables accurate automatic identification of IoT botnet variants equipped with new capabilities, which will help security researchers to investigate the new capabilities, and to apply the investigation findings for improving the solutions for detecting and preventing IoT botnet infections.