In this paper, we are going to verify the possibility to create a ransomware simulation that will use an arbitrary combination of known tactics and techniques to bypass an anti-malware defense. To verify this hypothesis, we conducted an experiment in which an agent was trained with the help of reinforcement learning to run the ransomware simulator in a way that can bypass anti-ransomware solution and encrypt the target files. The novelty of the proposed method lies in applying reinforcement learning to anti-ransomware testing that may help to identify weaknesses in the anti-ransomware defense and fix them before a real attack happens. © 2020 IEEE.

This paper contains an analysis of the LockerGoga ransomware that was used in the range of targeted cyberattacks in the first half of 2019 against Norsk Hydra-A world top 5 aluminum manufacturer, as well as the US chemical enterprises Hexion, and Momentive-Those companies are only the tip of the iceberg that reported the attack to the public. The ransomware was executed by attackers from inside a corporate network to encrypt the data on enterprise servers and, thus, taking down the information control systems. The intruders asked for a ransom to release a master key and decryption tool that can be used to decrypt the affected files. The purpose of the analysis is to find out tactics and techniques used by the LockerGoga ransomware during the cryptolocker attack as well as an encryption model to answer the question if the encrypted files can be decrypted with or without paying a ransom. The scientific novelty of the paper lies in an analysis methodology that is based on various reverse engineering techniques such as multi-process debugging and using open source code of a cryptographic library to find out a ransomware encryption model. © 2019 IEEE.

Engineering education in Sweden – as in the rest of the world – is experiencing a decline in student interest. There are concerns about the ways in which students think about engineering education, why they join an academic programme in engineering, and why they persist in their studies. In this context the aims of the Nationellt ämnesdidaktiskt Centrum för Teknikutbildning i Studenternas Sammanhang project (CeTUSS) is to investigate the student experience and to identify and support a continuing network of interested researchers, as well as in building capacity for disciplinary pedagogic investigation.

The Stepping Stones project brings together these interests in a multi-researcher, multi-institutional study that investigates how tudents and academic staff perceive engineering in Sweden and in Swedish education. The first results of that project are reported here. As this study is situated uniquely in Swedish education, it allows for exploration of “a Swedish perspective” on conceptions of engineering. The Stepping Stones project was based on a model of research capacity-building previously instantiated in the USA and Australia (Fincher & Tenenberg, 2006).

Different kinds of uncertainty, such as variations in manufacturing capability and functionality, as well as changes in demand, make up a dynamically changing environment for many manufacturing systems of today. The ability to adapt to these unforeseen changes, through dynamic decision-making as well as dynamic control capabilities based on the use of real-time manufacturing information and intelligence, is vital to be able to perform at a competitive level while reducing unscheduled downtime. The event-driven Function Block (FB) model of the IEC 61499 standard, as opposed to the time-triggered and data-driven concept of IEC 61331, supports this approach, making it possible to handle, in a responsive and adaptive way, different kinds of uncertainty. Our objective is to develop methodologies for distributed, adaptive and dynamic process planning as well as machine monitoring and control for machining and assembly operations, using event-driven FBs. The implementation and testing of FB-based control for manufacturing equipment has been successfully realized in prototype systems, with control of both CNC machining and robotic assembly operations. The approach of using IEC 61499 FBs for adaptive control in other applications is also investigated, as an adaptive decision support system for operators at manufacturing facilities is under development. We strongly believe that IEC 61499 will play a major role in the shift to adaptive manufacturing systems.

Partial evaluation is a program transformation technique that specializes a program with respect to part of its input. While the specialization is typically performed ahead-of-time, moving it to a later stage may expose additional opportunities and allow for faster residual programs to be constructed. In this thesis, we present a method for specializing programs at run-time, for compiled code, using an online partial evaluator. Although partial evaluation has several applications, the evaluation of the method primarily focuses on its performance benefits. The main research problem addressed in this thesis is that of incorporating an online partial evaluator in compiled code. The partial evaluator is a sourceto-source translator that takes and produces an abstract syntax tree (AST). Our approach consists of three parts, namely that of partially evaluating, obtaining a partially evaluable representation and run-time code emitting. Concretely, we use the concept of lifting to store an AST in the compiled code that the partial evaluator then specializes at run-time. The residual code is thereafter naively just-in-time (JIT) compiled through dynamically linking it back to the executable as a shared library. We evaluate the method on several programs and show that the specialized programs sometimes are faster even with a low recursion depth. Though, while the results are promising, the overhead is typically significant and therefore the break-even points are large. Further research, for example using an efficient JIT compiler, is required to better evaluate the performance benefits of the approach.

In this thesis we investigated performance of Kdtree, Vptree and Octree for storing neuronal morphology data. Two naive list structures were implemented to compare with the space partition data structures. The performance was measured with different sizes of neuronal networks and different types of test cases. A comparison with focus on cache misses, average search time and memory usage was made. Furthermore, measurements gathered quantitative data about each data structure. The results showed significant difference in performance of each data structure. It was concluded that Vptree is more suitable for searches in smaller populations of neurons and for specific nodes in larger populations, while Kdtree is better for volume searches in larger populations. Octree had highest average search time and memory requirement.

Starting from an empirical study of lay adults' understanding of heatand temperature, we distinguish between different meanings of "context" inphenomenographic research. To confuse the variation in ways of experiencingthe context(s) of the study with the variation in ways of experiencing thephenomenon of study is to risk losing fundamental insights. We discuss contextas experienced and as interwoven with the experience of the phenomenon, andanalyse its significance in two dimensions: (1) the stage of the research project:formulating the question, collecting data, analysing data and deploying results;and (2) "who is experiencing" the context: the individual, the collective, or theresearcher. The arguments are illustrated from the empirical study.

In this thesis, the impact of packet losses on the quality of received videos sent across a network that exhibit normal network perturbations such as jitters, delays, packet drops etc has been examined. Dynamic behavior of a normal network has been simulated using Linux and the Network Emulator (NetEm). Peoples’ perceptions on the quality of the received video were used in rating the qualities of several videos with differing speeds. In accordance with ITU’s guideline of using Mean Opinion Scores (MOS), the effects of packet drops were analyzed. Excel and Matlab were used as tools in analyzing the peoples’ opinions which indicates the impacts that different loss rates has on the transmitted videos. Statistical methods used for evaluation of data are mean and variance. We conclude that people have convergence of opinions when losses become extremely high on videos with highly variable scene changes

Cloud computing provides access to on-demand computing resources and storage space, whereby applications and data are hosted with data centers managed by third parties, on a pay-per-use price model. This allows organizations to focus on core business goals instead of managing in-house IT infrastructure.                    

However, as more business critical applications and data are moved to the cloud, service availability is becoming a growing concern. A number of recent cloud service disruptions have questioned the reliability of cloud environments to host business critical applications and data. The impact of these disruptions varies, but, in most cases, there are financial losses and damaged reputation among consumers.        

This thesis aims to investigate the threats to service availability in cloud computing and to provide some best practices to mitigate some of these threats. As a result, we identified eight categories of threats. They include, in no particular order: power outage, hardware failure, cyber-attack, configuration error, software bug, human error, administrative or legal dispute and network dependency. A number of systematic mitigation techniques to ensure constant availability of service by cloud providers were identified. In addition, practices that can be applied by cloud customers and users of cloud services, to improve service availability were presented.

As more companies add AI components to their software, most need to consider how these applications will evolve in the long term. The evolution of these AI components comes with risks, some of which come from the engineering practices adhered to when they are built. This is because traditional software practices do not apply when building AI applications. This creates technical debt, eventually impacting projects when project managers do not invest in mitigating them. This thesis discusses technical debts and describes the risks stemming from them. Mitigating these technical debts is a way to manage the risks to which these technical debts expose AI components and the larger system. Therefore, the research questions are: What risks from technical debt are associated with evolving AI components in software systems? What strategies are software companies adopting to assess risks associated with the evolution of AI components in software systems? And what approaches do software companies take to mitigate technical debt risks associated with the evolution of AI components in software systems? To answer these questions, we employed the qualitative analysis strategy. The data was collected by interviewing Software Engineers, Data Scientists, AI Engineers, Machine Learning Engineers, and Engineering Managers from various companies using predetermined questions. The analysis was done using the thematic analysis method. We found that the risks in the literature varied from those reported by our respondents. We also discovered two risk categories: system and business risk. We discovered that all respondents assess risk via monitoring and that risk mitigation varied slightly in literature from what our respondents reported. In conclusion, we corroborated the risks and mitigation strategies identified in the literature with our respondents, noting some differences. We learned that the primary assessment method used by all our respondents was monitoring and mitigation, which is crucial to prevent AI component failure.

African languages are spoken by over a billion people, but are underrepresented in NLP research and development. The challenges impeding progress include the limited availability of annotated datasets, as well as a lack of understanding of the settings where current methods are effective. In this paper, we make progress towards solutions for these challenges, focusing on the task of named entity recognition (NER). We create the largest human-annotated NER dataset for 20 African languages, and we study the behavior of state-of-the-art cross-lingual transfer methods in an Africa-centric setting, demonstrating that the choice of source language significantly affects performance. We show that choosing the best transfer language improves zero-shot F1 scores by an average of 14 points across 20 languages compared to using English. Our results highlight the need for benchmark datasets and models that cover typologically-diverse African languages.

Email-related incidents/crimes are on the rise owing to the fact that communication by electronic mail (e-mail) has become an important part of our daily lives. The technicality behind e-mail plays an important role when looking for digital evidence that can be used to create a hypothesis that can be used during litigation. During this process, it is needful to have a tool that can help to isolate email incidents as a potential crime scene in the wake of suspected attacks. The problem that this paper is addressing paper, is more centered on realizing an open-source email-forensic tool that used the header analysis approach. One advantage of this approach is that it helps investigators to collect digital evidence from e-mail systems, organize the collected data, analyze and discover any discrepancies in the header fields of an e-mail, and generates an evidence report. The main contribution of this paper focuses on generating a freshly computed hash that is attached to every generated report, to ensure the verifiability, reliability, and integrity of the reports to prove that they have not been modified in any way. Finally, this ensures that the sanctity and forensic soundness of the collected evidence are maintained. © 2023 ACM.

An organization's ability to succeed depends on the Confidentiality, Integrity, and Availability of its information. This implies that the organization's information and assets must be secured and protected. However, the regular occurrence of threats, risks, and intrusions could serve as a barrier to the security of this information. This has made the management of Information security a necessity. Organizations are then trying to be more agile by looking for ways to identify and embrace opportunities swiftly and confront these risks more quickly. Very little research has examined the relationships between Organizational Agility and Information Security. Hence, this study aims to investigate the management of Information Security in organizations while maintaining agility and highlighting the challenges encountered, and also addresses the research question: How do organizations manage information security while maintaining organizational agility?

The research strategy used is the Case Study, and the data collection methods used are semi-structured interviews and documents. The interview was conducted in a financial institution in Nigeria with seven security specialists, and documents were obtained from the company's website to help gain insights into the services and products offered. Thematic analysis was the data analysis method chosen. The findings revealed eighteen measures in which Information Security can be managed while maintaining Organizational Agility. Part of the identified measures are similar to those identified in previous research, while new measures are also discovered. Furthermore, these identified measures will be useful for other organizations, particularly financial institutions, to emulate in managing their Information Security and being agile while at it.

The development in smart meter technology has made grid operations more efficient based on fine-grained electricity usage data generated at different levels of time granularity. Consequently, machine learning algorithms have benefited from these data to produce useful models for important grid operations. Although machine learning algorithms need historical data to improve predictive performance, these data are not readily available for public utilization due to privacy issues. The existing smart grid data simulation frameworks generate grid data with implicit privacy concerns since the data are simulated from a few real energy consumptions that are publicly available. This paper addresses two issues in smart grid. First, it assesses the level of privacy violation with the individual household appliances based on synthetic household aggregate loads consumption. Second, based on the findings, it proposes two privacy-preserving mechanisms to reduce this risk. Three inference attacks are simulated and the results obtained confirm the efficacy of the proposed privacy-preserving mechanisms.

The introduction of advanced metering infrastructure (AMI) smart meters has given rise to fine-grained electricity usage data at different levels of time granularity. AMI collects high-frequency daily energy consumption data that enables utility companies and data aggregators to perform a rich set of grid operations such as demand response, grid monitoring, load forecasting and many more. However, the privacy concerns associated with daily energy consumption data has been raised. Existing studies on data anonymization for smart grid data focused on the direct application of perturbation algorithms, such as microaggregation, to protect the privacy of consumers. In this paper, we empirically show that reliance on microaggregation alone is not sufficient to protect smart grid data. Therefore, we propose DFTMicroagg algorithm that provides a dual level of perturbation to improve privacy. The algorithm leverages the benefits of discrete Fourier transform (DFT) and microaggregation to provide additional layer of protection. We evaluated our algorithm on two publicly available smart grid datasets with millions of smart meters readings. Experimental results based on clustering analysis using k-Means, classification via k-nearest neighbor (kNN) algorithm and mean hourly energy consumption forecast using Seasonal Auto-Regressive Integrated Moving Average with eXogenous (SARIMAX) factors model further proved the applicability of the proposed method. Our approach provides utility companies with more flexibility to control the level of protection for their published energy data.

Progress in the field of Non-Intrusive Load Monitoring (NILM) has been attributed to the rise in the application of artificial intelligence. Nevertheless, the ability of energy disaggregation algorithms to disaggregate different appliance signatures from aggregated smart grid data poses some privacy issues. This paper introduces a new notion of disclosure risk termed energy disaggregation risk. The performance of Sequence-to-Sequence (Seq2Seq) NILM deep learning algorithm along with three activation extraction methods are studied using two publicly available datasets. To understand the extent of disclosure, we study three inference attacks on aggregated data. The results show that Variance Sensitive Thresholding (VST) event detection method outperformed the other two methods in revealing households' lifestyles based on the signature of the appliances. To reduce energy disaggregation risk, we investigate the performance of two privacy-preserving mechanisms based on microaggregation and Discrete Fourier Transform (DFT). Empirically, for the first scenario of inference attack on UK-DALE, VST produces disaggregation risks of 99%, 100%, 89% and 99% for fridge, dish washer, microwave, and kettle respectively. For washing machine, Activation Time Extraction (ATE) method produces a disaggregation risk of 87%. We obtain similar results for other inference attack scenarios and the risk reduces using the two privacy-protection mechanisms.

The benefit associated with the deployment of Internet of Things (IoT) technology is increasing daily. IoT has revolutionized our ways of life, especially when we consider its applications in smart connected homes. Smart devices at home enable the collection of data from multiple sensors for a range of applications and services. Nevertheless, the security and privacy issues associated with aggregating multiple sensors’ data in smart connected homes have not yet been sufficiently prioritized. Along this development, this paper proposes HOMEFUS, a privacy and security-aware model that leverages information theoretic correlation analysis and gradient boosting to fuse multiple sensors’ data at the edge nodes of smart connected homes. HOMEFUS employs federated learning, edge and cloud computing to reduce privacy leakage of sensitive data. To demonstrate its applicability, we show that the proposed model meets the requirements for efficient data fusion pipelines. The model guides practitio ners and researchers on how to setup secure smart connected homes that comply with privacy laws, regulations, and standards. 

Internet of Things (IoT) technology has created a new dimension for data collection, transmission, processing, storage, and service delivery. With the advantages offered by IoT technologies, interest in smart home automation has increased over the years. Nevertheless, smart connected homes are characterized with the security and privacy problems that are associated with aggregating multiple sensors' data and exposing them to the Internet. In this paper, we propose LPM, a lightweight privacy-aware model that leverages information theoretic correlation analysis and gradient boosting to fuse multiple sensors' data at the edge nodes of smart connected homes. LPM employs federated learning, edge and cloud computing to reduce privacy leakages of sensitive data. To demonstrate its applicability, two services, commonly provided by smart homes, i.e., occupancy detection and people count estimation, were experimentally investigated. The results show that LPM can achieve accuracy, F1 score and AUC-ROC of 99.98%, 99.13%, and 99.98% respectively for occupancy detection as well as Mean Squared Error (MSE), Mean Absolute Error (MAE), and R2 of 0.0011,0.0175, and 98.39% respectively for people count estimation. LPM offers the opportunity to each smart connected home to participate in collaborative learning that is achieved through the federated machine learning component of the proposed model.

Progress in the field of Non-Intrusive Load Monitoring (NILM) has been attributed to the rise in the application of artificial intelligence. Nevertheless, the ability of energy disaggregation algorithms to disaggregate different appliance signatures from aggregated smart grid data poses some privacy issues. This paper introduces a new notion of disclosure risk termed energy disaggregation risk. The performance of Sequence-to-Sequence (Seq2Seq) NILM deep learning algorithm along with three activation extraction methods are studied using two publicly available datasets. To understand the extent of disclosure, we study three inference attacks on aggregated data. The results show that Variance Sensitive Thresholding (VST) event detection method outperformed the other two methods in revealing households' lifestyles based on the signature of the appliances. To reduce energy disaggregation risk, we investigate the performance of two privacy-preserving mechanisms based on microaggregation and Discrete Fourier Transform (DFT). Empirically, for the first scenario of inference attack on UK-DALE, VST produces disaggregation risks of 99%, 100%, 89% and 99% for fridge, dish washer, microwave, and kettle respectively. For washing machine, Activation Time Extraction (ATE) method produces a disaggregation risk of 87%. We obtain similar results for other inference attack scenarios and the risk reduces using the two privacy-protection mechanisms.

The advancement in artificial intelligence (AI) techniques has given rise to the success rate recorded in the field of Non-Intrusive Load Monitoring (NILM). The development of robust AI and machine learning algorithms based on deep learning architecture has enabled accurate extraction of individual appliance load signature from aggregated energy data. However, the success rate of NILM algorithm in disaggregating individual appliance load signature in smart grid data violates the privacy of the individual household lifestyle. This paper investigates the performance of Sequence-to-Sequence (Seq2Seq) deep learning NILM algorithm in predicting the load signature of appliances. Furthermore, we define a new notion of disclosure risk to understand the risk associated with individual appliances in aggregated signals. Two publicly available energy disaggregation datasets have been considered. We simulate three inference attack scenarios to better ascertain the risk of publishing raw energy data. In addition, we investigate three activation extraction methods for appliance event detection. The results show that the disclosure risk associated with releasing smart grid data in their original form is on the high side. Therefore, future privacy protection mechanisms should devise efficient methods to reduce this risk.

The development in smart meter technology has made grid operations more efficient based on fine-grained electricity usage data generated at different levels of time granularity. Consequently, machine learning algorithms have benefited from these data to produce useful models for important grid operations. Although machine learning algorithms need historical data to improve predictive performance, these data are not readily available for public utilization due to privacy issues. The existing smart grid data simulation frameworks generate grid data with implicit privacy concerns since the data are simulated from a few real energy consumptions that are publicly available. This paper addresses two issues in smart grid. First, it assesses the level of privacy violation with the individual household appliances based on synthetic household aggregate loads consumption. Second, based on the findings, it proposes two privacy-preserving mechanisms to reduce this risk. Three inference attacks are simulated and the results obtained confirm the efficacy of the proposed privacy-preserving mechanisms.

This thesis presents resources capable of enhancing solutions of some Natural Language Processing (NLP) tasks, demonstrates the learning of abstractions by deep models through cross-lingual transferability, and shows how deep learning models trained on idioms can enhance open-domain conversational systems. The challenges of open-domain conversational systems are many and include bland repetitive utterances, lack of utterance diversity, lack of training data for low-resource languages, shallow world-knowledge and non-empathetic responses, among others. These challenges contribute to the non-human-like utterances that open-domain conversational systems suffer from. They, hence,have motivated the active research in Natural Language Understanding (NLU) and Natural Language Generation (NLG), considering the very important role conversations (or dialogues) play in human lives. The methodology employed in this thesis involves an iterative set of scientific methods. First, it conducts a systematic literature review to identify the state-of-the-art (SoTA) and gaps, such as the challenges mentioned earlier, in current research. Subsequently, it follows the seven stages of the Machine Learning (ML) life-cycle, which are data gathering (or acquisition), data preparation, model selection, training, evaluation with hyperparameter tuning, prediction and model deployment. For data acquisition, relevant datasets are acquired or created, using benchmark datasets as references, and their data statements are included. Specific contributions of this thesis are the creation of the Swedish analogy test set for evaluating word embeddings and the Potential Idiomatic Expression (PIE)-English idioms corpus for training models in idiom identification and classification. In order to create a benchmark, this thesis performs human evaluation on the generated predictions of some SoTA ML models, including DialoGPT. As different individuals may not agree on all the predictions, the Inter-Annotator Agreement (IAA) is measured. A typical method for measuring IAA is Fleiss Kappa, however, it has a number of shortcomings, including high sensitivity to the number of categories being evaluated. Therefore, this thesis introduces the credibility unanimous score (CUS), which is more intuitive, easier to calculate and seemingly less sensitive to changes in the number of categories being evaluated. The results of human evaluation and comments from evaluators provide valuable feedback on the existing challenges within the models. These create the opportunity for addressing such challenges in future work. The experiments in this thesis test two hypothesis; 1) an open-domain conversational system that is idiom-aware generates more fitting responses to prompts containing idioms, and 2) deep monolingual models learn some abstractions that generalise across languages. To investigate the first hypothesis, this thesis trains English models on the PIE-English idioms corpus for classification and generation. For the second hypothesis, it explores cross-lingual transferability from English models to Swedish, Yorùbá, Swahili, Wolof, Hausa, Nigerian Pidgin English and Kinyarwanda. From the results, the thesis’ additional contributions mainly lie in 1) confirmation of the hypothesis that an open-domain conversational system that is idiom-aware generates more fitting responses to prompts containing idioms, 2) confirmation of the hypothesis that deep monolingual models learn some abstractions that generalise across languages, 3) introduction of CUS and its benefits, 4) insight into the energy-saving and time-saving benefits of more optimal embeddings from relatively smaller corpora, and 5) provision of public access to the model checkpoints that were developed from this work. We further discuss the ethical issues involved in developing robust, open-domain conversational systems. Parts of this thesis are already published in the form of peer-reviewed journal and conference articles.

Building open-domain conversational systems (or chatbots) that produce convincing responses is a recognized challenge. Recent state-of-the-art (SoTA) transformer-based models for the generation of natural language dialogue have demonstrated impressive performance in simulating human-like, single-turn conversations in English.This work investigates, by an empirical study, the potential for transfer learning of such models to Swedish language. DialoGPT, an English language pre-trained model, is adapted by training on three different Swedish language conversational datasets obtained from publicly available sources: Reddit, Familjeliv and the GDC. Perplexity score (an automated intrinsic metric) and surveys by human evaluation were used to assess the performances of the fine-tuned models. We also compare the DialoGPT experiments with an attention-mechanism-based seq2seq baseline model, trained on the GDC dataset. The results indicate that the capacity for transfer learning can be exploited with considerable success. Human evaluators asked to score the simulated dialogues judged over 57% of the chatbot responses to be human-like for the model trained on the largest (Swedish) dataset. The work agrees with the hypothesis that deep monolingual models learn some abstractions which generalize across languages. We contribute the codes, datasets and model checkpoints and host the demos on the HuggingFace platform.

Building open-domain conversational systems (or chatbots) that produce convincing responses is a recognized challenge. Recent state-of-the-art (SoTA) transformer-based models for the generation of natural language dialogue have demonstrated impressive performance in simulating human-like, single-turn conversations in English.This work investigates, by an empirical study, the potential for transfer learning of such models to Swedish language. DialoGPT, an English language pre-trained model, is adapted by training on three different Swedish language conversational datasets obtained from publicly available sources: Reddit, Familjeliv and the GDC. Perplexity score (an automated intrinsic metric) and surveys by human evaluation were used to assess the performances of the fine-tuned models. We also compare the DialoGPT experiments with an attention-mechanism-based seq2seq baseline model, trained on the GDC dataset. The results indicate that the capacity for transfer learning can be exploited with considerable success. Human evaluators asked to score the simulated dialogues judged over 57% of the chatbot responses to be human-like for the model trained on the largest (Swedish) dataset. The work agrees with the hypothesis that deep monolingual models learn some abstractions which generalize across languages. We contribute the codes, datasets and model checkpoints and host the demos on the HuggingFace platform.

This essay discusses current research efforts in conversational systems from the philosophy of science point of view and evaluates some conversational systems research activities from the standpoint of naturalism philosophical theory. Conversational systems or chatbots have advanced over the decades and now have become mainstream applications. They are software that users can communicate with, using natural language. Particular attention is given to the Alime Chat conversational system, already in industrial use, and the related research. The competitive nature of systems in production is a result of different researchers and developers trying to produce new conversational systems that can outperform previous or state-of-the-art systems. Different factors affect the quality of the conversational systems produced, and how one system is assessed as being better than another is a function of objectivity and of the relevant experimental results. This essay examines the research practices from, among others, Longino’s view on objectivity and Popper’s stand on falsification. Furthermore, the need for qualitative and large datasets is emphasized. This is in addition to the importance of the peer-review process in scientific publishing, as a means of developing, validating, or rejecting theories, claims, or methodologies in the research community. In conclusion, open data and open scientific discussion fora should become more prominent over the mere publication-focused trend.

In this work, we show that the difference in performance of embeddings from differently sourced data for a given language can be due to other factors besides data size. Natural language processing (NLP) tasks usually perform better with embeddings from bigger corpora. However, broadness of covered domain and noise can play important roles. We evaluate embeddings based on two Swedish corpora: The Gigaword and Wikipedia, in analogy (intrinsic) tests and discover that the embeddings from the Wikipedia corpus generally outperform those from the Gigaword corpus, which is a bigger corpus. Downstream tests will be required to have a definite evaluation.

In this paper, the authors propose to increase the efficiency of blockchain mining by using a population-based approach. Blockchain relies on solving difficult mathematical problems as proof-of-work within a network before blocks are added to the chain. Brute force approach, advocated by some as the fastest algorithm for solving partial hash collisions and implemented in Bitcoin blockchain, implies exhaustive, sequential search. It involves incrementing the nonce (number) of the header by one, then taking a double SHA-256 hash at each instance and comparing it with a target value to ascertain if lower than that target. It excessively consumes both time and power. In this paper, the authors, therefore, suggest using an inner for-loop for the population-based approach. Comparison shows that it’s a slightly faster approach than brute force, with an average speed advantage of about 1.67% or 3,420 iterations per second and 73% of the time performing better. Also, we observed that the more the total particles deployed, the better the performance until a pivotal point. Furthermore, a recommendation on taming the excessive use of power by networks, like Bitcoin’s, by using penalty by consensus is suggested.

We conduct relatively extensive investigations of automatic hate speech (HS) detection using different State-of-The-Art (SoTA) baselines across 11 subtasks spanning six different datasets. Our motivation is to determine which of the recent SoTA models is best for automatic hate speech detection and what advantage methods, such as data augmentation and ensemble, may have on the best model, if any. We carry out six cross-task investigations. We achieve new SoTA results on two subtasks—macro F1 scores of 91.73% and 53.21% for subtasks A and B of the HASOC 2020 dataset, surpassing previous SoTA scores of 51.52% and 26.52%, respectively. We achieve near-SoTA results on two others—macro F1 scores of 81.66% for subtask A of the OLID 2019 and 82.54% for subtask A of the HASOC 2021, in comparison to SoTA results of 82.9% and 83.05%, respectively. We perform error analysis and use two eXplainable Artificial Intelligence (XAI) algorithms (Integrated Gradient (IG) and SHapley Additive exPlanations (SHAP)) to reveal how two of the models (Bi-Directional Long Short-Term Memory Network (Bi-LSTM) and Text-to-Text-Transfer Transformer (T5)) make the predictions they do by using examples. Other contributions of this work are: (1) the introduction of a simple, novel mechanism for correcting Out-of-Class (OoC) predictions in T5, (2) a detailed description of the data augmentation methods, and (3) the revelation of the poor data annotations in the HASOC 2021 dataset by using several examples and XAI (buttressing the need for better quality control). We publicly release our model checkpoints and codes to foster transparency.

Dialogue generation is an important NLP task fraught with many challenges. The challenges become more daunting for low-resource African languages. To enable the creation of dialogue agents for African languages, we contribute the first high-quality dialogue datasets for 6 African languages: Swahili, Wolof, Hausa, Nigerian Pidgin English, Kinyarwanda & Yorùbá. There are a total of 9,000 turns, each language having 1,500 turns, which we translate from a portion of the English multi-domain MultiWOZ dataset. Subsequently, we benchmark by investigating & analyzing the effectiveness of modelling through transfer learning by utilziing state-of-the-art (SoTA) deep monolingual models: DialoGPT and BlenderBot. We compare the models with a simple seq2seq baseline using perplexity. Besides this, we conduct human evaluation of single-turn conversations by using majority votes and measure inter-annotator agreement (IAA). We find that the hypothesis that deep monolingual models learn some abstractions that generalize across languages holds. We observe human-like conversations, to different degrees, in 5 out of the 6 languages. The language with the most transferable properties is the Nigerian Pidgin English, with a human-likeness score of 78.1%, of which 34.4% are unanimous. We freely provide the datasets and host the model checkpoints/demos on the HuggingFace hub for public access.

We introduce the Instruction Document Visual Question Answering (iDocVQA) dataset and the Large Language Document (LLaDoc) model, for training Language-Vision (LV) models for document analysis and predictions on document images, respectively. Usually, deep neural networks for the DocVQA task are trained on datasets lacking instructions. We show that using instruction-following datasets improves performance. We compare performance across document-related datasets using the recent state-of-the-art (SotA) Large Language and Vision Assistant (LLaVA)1.5 as the base model. We also evaluate the performance of the derived models for object hallucination using the Polling-based Object Probing Evaluation (POPE) dataset. The results show that instruction-tuning performance ranges from 11x to 32x of zero-shot performance and from 0.1% to 4.2% over non-instruction (traditional task) finetuning. Despite the gains, these still fall short of human performance (94.36%), implying there’s much room for improvement.

We survey SoTA open-domain conversational AI models with the objective of presenting the prevailing challenges that still exist to spur future research. In addition, we provide statistics on the gender of conversational AI in order to guide the ethics discussion surrounding the issue. Open-domain conversational AI models are known to have several challenges, including bland, repetitive responses and performance degradation when prompted with figurative language, among others. First, we provide some background by discussing some topics of interest in conversational AI. We then discuss the method applied to the two investigations carried out that make up this study. The first investigation involves a search for recent SoTA open-domain conversational AI models, while the second involves the search for 100 conversational AI to assess their gender. Results of the survey show that progress has been made with recent SoTA conversational AI, but there are still persistent challenges that need to be solved, and the female gender is more common than the male for conversational AI. One main takeaway is that hybrid models of conversational AI offer more advantages than any single architecture. The key contributions of this survey are (1) the identification of prevailing challenges in SoTA open-domain conversational AI, (2) the rarely held discussion on open-domain conversational AI for low-resource languages, and (3) the discussion about the ethics surrounding the gender of conversational AI.

The major contributions of this work include the empirical establishment of a better performance for Yoruba embeddings from undiacritized (normalized) dataset and provision of new analogy sets for evaluation.The Yoruba language, being a tonal language, utilizes diacritics (tonal marks) in written form. We show that this affects embedding performance by creating embeddings from exactly the same Wikipedia dataset but with the second one normalized to be undiacritized. We further compare average intrinsic performance with two other work (using analogy test set & WordSim) and we obtain the best performance in WordSim and corresponding Spearman correlation.

Open-domain chatbots have advanced but still have many gaps. My PhD aims to solve a few of those gaps by creating vector representations of idioms (figures of speech) that will be beneficial to chatbots and natural language processing (NLP), generally. In the process, new, optimal fastText embeddings in Swedish and English have been created and the first Swedish analogy test set, larger than the Google original, for intrinsic evaluation of Swedish embeddings has also been produced. Major milestones have been attained and others are soon to follow. The deliverables of this project will give NLP researchers the opportunity to measure the quality of Swedish embeddings easily and advance state-of-the-art (SotA) in NLP.

The purpose of this study is to present guidelines that can be followed when introducing Service-oriented architecture through the use of Web services. This guideline will be especially useful for organizations migrating from their existing legacy systems where the need also arises to consider the financial implications of such an investment whether it is worthwhile or not. The proposed implementation guide aims at increasing the chances of IT departments in organizations to ensure a successful integration of SOA into their system and secure strong financial commitment from the executive management. Service oriented architecture technology is a new concept, a new way of looking at a system which has emerged in the IT world and can be implemented by several methods of which Web services is one platform. Since it is a developing technology, organizations need to be cautious on how to implement this technology to obtain maximum benefits. Though a well-designed, service-oriented environment can simplify and streamline many aspects of information technology and business, achieving this state is not an easy task. Traditionally, management finds it very difficult to justify the considerable cost of modernization, let alone shouldering the risk without achieving some benefits in terms of business value. The study identifies some common best practices of implementing SOA and the use of Web services, steps to successfully migrate from legacy systems to componentized or service enabled systems. The study also identified how to present financial return on investment and business benefits to the management in order to secure the necessary funds. This master thesis is based on academic literature study, professional research journals and publications, interview with business organizations currently working on service oriented architecture. I present guidelines that can be of assistance to migrate from legacy systems to service-oriented architecture based on the analysis from comparing information sources mentioned above.

Among the more salient accelerator technologies to continue performance scaling in High-Performance Computing (HPC) are Coarse-Grained Reconfigurable Arrays (CGRAs). However, what benefits CGRAs will bring to HPC workloads and how those benefits will be reaped is an open research question today. In this work, we propose a framework to explore the design space of CGRAs for HPC workloads, which includes a tool flow of compilation and simulation, a CGRA HDL library written in SystemVerilog, and a synthesizable CGRA design as a baseline. Using RTL simulation, we evaluate two well-known computation kernels with the baseline CGRA for multiple different architectural parameters. The simulation results demonstrate both correctness and usefulness of our exploration framework.

This research aims to explore the tradeoffs between routing flexibility and hardware resource usage, ultimately reducing the resource usage of our CGRA architecture while maintaining compute efficiency. we investigate statistics of connection usages among switch blocks for benchmark DFGs, propose several CGRA architecture with a reduced connection, and evaluate their hardware cost, routability of DFGs, and computational throughput for benchmarks. We found that the topology with horizontal plus diagonal connection saves about 30% of the resource usage while maintaining virtually the same routing flexibility as the full connectivity topology.

Smart grid systems (SGSs), and in particular power systems, play a vital role in today's urban life. The security of these grids is now threatened by adversaries that use false data injection (FDI) to produce a breach of availability, integrity, or confidential principles of the system. We propose a novel structure for the multigenerator generative adversarial network (GAN) to address the challenges of detecting adversarial attacks. We modify the GAN objective function and the training procedure for the malicious anomaly detection task. The model only requires normal operation data to be trained, making it cheaper to deploy and robust against unseen attacks. Moreover, the model operates on the raw input data, eliminating the need for feature extraction. We show that the model reduces the well-known mode collapse problem of GAN-based systems, it has low computational complexity and considerably outperforms the baseline system (OCAN) with about 55% in terms of accuracy on a freely available cyber attack dataset.

We propose a multi-layer variational autoencoder method, we call HR-VQVAE, thatlearns hierarchical discrete representations of the data. By utilizing a novel objectivefunction, each layer in HR-VQVAE learns a discrete representation of the residual fromprevious layers through a vector quantized encoder. Furthermore, the representations ateach layer are hierarchically linked to those at previous layers. We evaluate our methodon the tasks of image reconstruction and generation. Experimental results demonstratethat the discrete representations learned by HR-VQVAE enable the decoder to reconstructhigh-quality images with less distortion than the baseline methods, namely VQVAE andVQVAE-2. HR-VQVAE can also generate high-quality and diverse images that outperform state-of-the-art generative models, providing further verification of the efficiency ofthe learned representations. The hierarchical nature of HR-VQVAE i) reduces the decoding search time, making the method particularly suitable for high-load tasks and ii) allowsto increase the codebook size without incurring the codebook collapse problem.

Cyber attacks and anomaly detection are problems where the data is often highly unbalanced towards normal observations. Furthermore, the anomalies observed in real applications may be significantly different from the ones contained in the training data. It is, therefore, desirable to study methods that are able to detect anomalies only based on the distribution of the normal data. To address this problem, we propose a novel objective function for generative adversarial networks (GANs), referred to as STEPGAN. STEP-GAN simulates the distribution of possible anomalies by learning a modified version of the distribution of the task-specific normal data. It leverages multiple generators in a step-by-step interaction with a discriminator in order to capture different modes in the data distribution. The discriminator is optimized to distinguish not only between normal data and anomalies but also between the different generators, thus encouraging each generator to model a different mode in the distribution. This reduces the well-known mode collapse problem in GAN models considerably. We tested our method in the areas of power systems and network traffic control systems (NTCSs) using two publicly available highly imbalanced datasets, ICS (Industrial Control System) security dataset and UNSW-NB15, respectively. In both application domains, STEP-GAN outperforms the state-of-the-art systems as well as the two baseline systems we implemented as a comparison. In order to assess the generality of our model, additional experiments were carried out on seven real-world numerical datasets for anomaly detection in a variety of domains. In all datasets, the number of normal samples is significantly more than that of abnormal samples. Experimental results show that STEP-GAN outperforms several semi-supervised methods while being competitive with supervised methods.

Collaborative artificial intelligence systems (CAISs) aim to work with humans in a shared space to achieve a common goal, but this can pose hazards that could harm human beings. We identify emerging problems in this context and report our vision of and progress toward a risk-driven assurance process for CAISs.

Non-repudiation means that a person is unable to deny a certain action that he has done under any circumstances. There are several mechanisms, standards and protocols to achieve non-repudiation in techno-information enviromnent. However efficiency in non-repudiation in legal framework was not considerably addressed within the context of those mechanisms. Lack of efficient non-repudiation in the legal framework for techno-information environment makes legal issues when evidence is generated maintained. It can be derived that traditional non-repudiation mechanism delivers efficient non-repudiation. Efficient non-repudiation in techno-information environment is achieved by mapping traditional non-repudiation. Evaluation methodology for efficiency of non-repudiation mechanisms has been improved during this work. Further most significant finding of this research is the Efficient Non-Repudiation Protocol.

The existing approaches to design efficient safety critical control applications is constrained by limited in-vehicle sensing and computational capabilities. In the context of automated driving, we argue that there is a need to leverage resources "out-of-the-vehicle" to meet the sensing and powerful processing requirements of sophisticated algorithms (e.g., deep neural networks). To realize the need, a suitable computation offloading technique that meets the vehicle safety and stability requirements, even in the presence of unreliable communication network, has to be identified. In this work, we propose an adaptive offloading technique for control computations into the cloud. The proposed approach considers both current network conditions and control application requirements to determine the feasibility of leveraging remote computation and storage resources. As a case study, we describe a cloud-based path following controller application that leverages crowdsensed data for path planning.

This thesis explores the application of Reinforcement Learning (RL) to production job scheduling, specifically within the context of a factory environment. We developed an environment reflecting the dynamic operations of a factory floor, incorporating machines, jobs, and operations to be performed on jobs called recipes. We train Deep Q-Network (DQN) and Advantage Actor Critic (A2C) RL agents with function approximation using neural networks (NN) since the state space is continuous.

An important aspect of our work is the development of an RL agent capable of directly assigning jobs to machines without relying on dispatch rules. Instead, the actions taken by the agent are job-machine pairings. Throughout the training process, the environment, observation space, and reward structures were incrementally refined based on agent performance. The primary objective is to minimize the average number of jobs completed late, termed as tardiness. Our results demonstrate that RL agents can learn to make effective scheduling decisions, showing promising improvements in minimizing total jobs completed late compared to the traditional dispatch rules.

Shell scripting languages such as bash are designed to integrate with an OS, which mainly involves managing processes with implicit input and output streams. They also attempt to do this in a compact way that could be reasonably typed on a command-line interface. However, existing shell languages are not sufficient to serve as general-purpose languages-values are not observable except in raw streams of bytes, and they lack modern language features such as lexical scope and higher-order functions. By way of a new programming language, Magritte, we propose a general-purpose programming language with semantics similar to bash. In this paper, we discuss the early design of such a system, in which the primary unit of composition, like bash, is processes with input and output channels, which can be read from or written to at any time, and which can be chained together via a pipe operator. We also explore concurrency semantics for such a language.

The paper considers the problem of performing a post-processing task defined on a model parameter that is only observed indirectly through noisy data in an ill-posed inverse problem. A key aspect is to formalize the steps of reconstruction and post-processing as appropriate estimators (non-randomized decision rules) in statistical estimation problems. The implementation makes use of (deep) neural networks to provide a differentiable parametrization of the family of estimators for both steps. These networks are combined and jointly trained against suitable supervised training data in order to minimize a joint differentiable loss function, resulting in an end-to-end task adapted reconstruction method. The suggested framework is generic, yet adaptable, with a plug-and-play structure for adjusting both the inverse problem and the post-processing task at hand. More precisely, the data model (forward operator and statistical model of the noise) associated with the inverse problem is exchangeable, e.g., by using neural network architecture given by a learned iterative method. Furthermore, any post-processing that can be encoded as a trainable neural network can be used. The approach is demonstrated on joint tomographic image reconstruction, classification and joint tomographic image reconstruction segmentation.

Gaze tracking has already shown to be a popular technology for desktop devices. When it comes to gaze tracking for mobile devices, however, there is still a lot of progress to be made. There’s still no high accuracy gaze tracking available that works in an unconstrained setting for mobile devices. This work makes contributions in the area of appearance-based unconstrained gaze estimation. Artificial neural networks are trained on GazeCapture, a publicly available dataset for mobile gaze estimation containing over 2 million face images and corresponding gaze labels. In this work, Siamese neural networks are trained to learn linear distances between face images for different gaze points. Then, during inference, calibration points are used to estimate gaze points. This approach is shown to be an effective way of utilizing calibration points in order to improve the result of gaze estimation.

We present and discuss algorithms and library software for solving the generalized non-symmetric eigenvalue problem (GNEP) on high performance computing (HPC) platforms with distributed memory. Such problems occur frequently in computational science and engineering, and our contributions make it possible to solve GNEPs fast and accurate in parallel using state-of-the-art HPC systems. A generalized eigenvalue problem corresponds to finding scalars y and vectors x such that Ax = yBx, where A and B are real square matrices. A nonzero x that satisfies the GNEP equation is called an eigenvector of the ordered pair (A,B), and the scalar y is the associated eigenvalue. Our contributions include parallel algorithms for transforming a matrix pair (A,B) to a generalized Schur form (S,T), where S is quasi upper triangular and T is upper triangular. The eigenvalues are revealed from the diagonals of S and T. Moreover, for a specified set of eigenvalues an associated pair of deflating subspaces can be computed, which typically is requested in various applications. In the first stage the matrix pair (A,B) is reduced to a Hessenberg-triangular form (H,T), where H is upper triangular with one nonzero subdiagonal and T is upper triangular, in a finite number of steps. The second stage reduces the matrix pair further to generalized Schur form (S,T) using an iterative QZ-based method. Outgoing from a one-stage method for the reduction from (A,B) to (H,T), a novel parallel algorithm is developed. In brief, a delayed update technique is applied to several partial steps, involving low level operations, before associated accumulated transformations are applied in a blocked fashion which together with a wave-front task scheduler makes the algorithm scale when running in a parallel setting. The potential presence of infinite eigenvalues makes a generalized eigenvalue problem ill-conditioned. Therefore the parallel algorithm for the second stage, reduction to (S,T) form, continuously scan for and robustly deflate infinite eigenvalues. This will reduce the impact so that they do not interfere with other real eigenvalues or are misinterpreted as real eigenvalues. In addition, our parallel iterative QZ-based algorithm makes use of multiple implicit shifts and an aggressive early deflation (AED) technique, which radically speeds up the convergence. The multi-shift strategy is based on independent chains of so called coupled bulges and computational windows which is an important source of making the algorithm scalable. The parallel algorithms have been implemented in state-of-the-art library software. The performance is demonstrated and evaluated using up to 1600 CPU cores for problems with matrices as large as 100000 x 100000. Our library software is described in a User Guide. The software is, optionally, tunable via a set of parameters for various thresholds and buffer sizes etc. These parameters are discussed, and recommended values are specified which should result in reasonable performance on HPC systems similar to the ones we have been running on.

In this paper we discuss the problem of computing eigenvectors for matrices in Schur form using parallel computing. We develop a new parallel algorithm and report on the performance of our MPI based implementation. We have also implemented a new parallel algorithm for scaling during the backsubstitution phase. We have increased the arithmetic intensity by interleaving the compution of several eigenvectors and by merging the backward substitution and the back-transformation of the eigenvector computation.