We participated in the WMT 2022 Large-Scale Machine Translation Evaluation for the African Languages Shared Task. This work de-scribes our approach, which is based on filtering the given noisy data using a sentence-pair classifier that was built by fine-tuning a pre-trained language model. To train the classifier, we obtain positive samples (i.e. high-quality parallel sentences) from a gold-standard curated dataset and extract negative samples (i.e.low-quality parallel sentences) from automatically aligned parallel data by choosing sentences with low alignment scores. Our final machine translation model was then trained on filtered data, instead of the entire noisy dataset. We empirically validate our approach by evaluating on two common datasets and show that data filtering generally improves overall translation quality, in some cases even significantly.

The thesis focuses on machine learning methods for Earth Observation (EO) data, more specifically, remote sensing data acquired by satellites and drones. EO plays a vital role in monitoring the Earth’s surface and modelling climate change to take necessary precautionary measures. Initially, these efforts were dominated by methods relying on handcrafted features and expert knowledge. The recent advances of machine learning methods, however, have also led to successful applications in EO. This thesis explores supervised and unsupervised approaches of Deep Learning (DL) to monitor natural resources of water bodies and forests. 

The first study of this thesis introduces an Unsupervised Curriculum Learning (UCL) method based on widely-used DL models to classify water resources from RGB remote sensing imagery. In traditional settings, human experts labeled images to train the deep models which is costly and time-consuming. UCL, instead, can learn the features progressively in an unsupervised fashion from the data, reducing the exhausting efforts of labeling. Three datasets of varying resolution are used to evaluate UCL and show its effectiveness: SAT-6, EuroSAT, and PakSAT. UCL outperforms the supervised methods in domain adaptation, which demonstrates the effectiveness of the proposed algorithm. 

The subsequent study is an extension of UCL for the multispectral imagery of Australian wildfires. This study has used multispectral Sentinel-2 imagery to create the dataset for the forest fires ravaging Australia in late 2019 and early 2020. 12 out of the 13 spectral bands of Sentinel-2 are concatenated in a way to make them suitable as a three-channel input to the unsupervised architecture. The unsupervised model then classified the patches as either burnt or not burnt. This work attains 87% F1-Score mapping the burnt regions of Australia, demonstrating the effectiveness of the proposed method. 

The main contributions of this work are (i) the creation of two datasets using Sentinel-2 Imagery, PakSAT dataset and Australian Forest Fire dataset; (ii) the introduction of UCL that learns the features progressively without the need of labelled data; and (iii) experimentation on relevant datasets for water body and forest fire classification. 

This work focuses on patch-level classification which could in future be expanded to pixel-based classification. Moreover, the methods proposed in this study can be extended to the multi-class classification of aerial imagery. Further possible future directions include the combination of geo-referenced meteorological and remotely sensed image data to explore proposed methods. Lastly, the proposed method can also be adapted to other domains involving multi-spectral and multi-modal input, such as, historical documents analysis, forgery detection in documents, and Natural Language Processing (NLP) classification tasks.

Earth Observation (EO) data is crucial for understanding, managing, and conserving our planet's ecosystem and its natural resources. This data enables humanity to monitor environmental changes, such as natural disasters, urban growth, and climate shifts, assisting informed decisions and proactive measures. Early EO heavily relied on statistical methods and expert domain knowledge, but the advent of machine learning has revolutionized EO data processing, enhancing efficiency and accuracy. Conventional ML models require expensive and labor-intensive data labeling. In contrast, unsupervised ML techniques can learn features from data without the need for manual labeling, making the process more efficient and cost-effective.

This thesis presents a UCL approach utilizing advanced DL models to classify EO data, referred to as UCL4EO. This approach eliminates the need for manual data labeling in training the DL model. The UCL framework comprises i) a DL model tailored for feature extraction from image data, ii) a clustering method to group deep features, and iii) a selection operation to capture representative samples from these clusters. The CNN extracts meaningful features from images, subjected to a clustering algorithm to create pseudo-labels. After identifying the initial clusters, representative samples from each cluster are chosen using the UCL selection operation to fine-tune the feature extractor. The stated process is repeated iteratively until convergence. The proposed UCL approach progressively learns and incorporates salient data features in an unsupervised manner by utilizing pseudo-labels.

UCL started as a proof of concept to show the viability of the method for binary classification on RS and aerial imagery. Specifically, the UCL framework is employed to identify water bodies using three RGB datasets, encompassing both low and high-resolution RS and aerial imagery. While UCL has been extensively examined with RGB imagery, it has been adapted to benefit from the enhanced capabilities of multi-spectral satellite imagery. This adaptation enables UCL to generalize to multi-spectral imagery from Sentinel-2 to detect forest fires in Australia. UCL undergoes subsequent improvements and is further investigated to identify utility poles in high-resolution UAV images. These gray-scale images of utility poles pose computer vision challenges, including issues like occlusion and cropping, where a significant portion of the image contains the background and only a slight appearance of the utility pole. Extensive experimentation on the mentioned tasks effectively showcases UCL's adaptive learning capabilities, producing promising results. The achieved accuracy surpassed those of supervised methods in cross-domain adaptation on similar tasks, underscoring the effectiveness of the proposed algorithm.

The scope of UCL has been extended to encompass multi-class classification tasks in the domain of RS data, referred to as Multi-class UCL. Multi-class UCL progressively acquires knowledge about various categories on multi-scale resolution. To investigate Multi-class UCL, we have used four publicly available datasets of Sentinel-2 and aerial imagery: EuroSAT, SAT-6, UCMerced, and RSSCN7. Comprehensive experiments conducted on the above-mentioned datasets revealed better cross-domain adaptation capabilities compared to supervised methods, thereby demonstrating the effectiveness of Multi-class UCL.

In these investigations, two datasets are generated using Sentinel-2 satellite imagery: one for water bodies - PakSAT and the other for Australian forest fires. However, cloud cover poses a significant challenge by obstructing the satellite's ability to capture clear images of the Earth's surface. To address this issue, available cloud masking techniques are employed to filter out images affected by cloud cover, ensuring the datasets contain only clear and usable data. Later, this thesis examines cloud detection and Cloud Optical Thickness (COT) estimation from Sentinel-2 imagery. We employed machine-learning techniques, achieving better performance than SCL designed by ESA for cloud cover tasks.

In addition to the application in RS data, UCL has been investigated in other domains of EO, such as undersea imagery. Furthermore, UCL has also been used for tasks like natural scene classification, medical imaging, and document analysis, demonstrating its versatility and broad applicability. Further exploration of UCL could involve improving the process of generating pseudo-labels through deep learning techniques.

This paper introduces a machine learning-based approach for detecting electric poles, an essential part of power grid maintenance. With the increasing popularity of deep learning, several such approaches have been proposed for electric pole detection. However, most of these approaches are supervised, requiring a large amount of labeled data, which is time-consuming and labor-intensive. Unsupervised deep learning approaches have the potential to overcome the need for huge amounts of training data. This paper presents an unsupervised deep learning framework for utility pole detection. The framework combines Convolutional Neural Network (CNN) and clustering algorithms with a selection operation. The CNN architecture for extracting meaningful features from aerial imagery, a clustering algorithm for generating pseudo labels for the resulting features, and a selection operation to filter out reliable samples to fine-tune the CNN architecture further. The fine-tuned version then replaces the initial CNN model, thus improving the framework, and we iteratively repeat this process so that the model learns the prominent patterns in the data progressively. The presented framework is trained and tested on a small dataset of utility poles provided by “Mention Fuvex” (a Spanish company utilizing long-range drones for power line inspection). Our extensive experimentation demonstrates the progressive learning behavior of the proposed method and results in promising classification scores with significance test having p−value<0.00005 on the utility pole dataset.

Massive wildfires not only in Australia, but also worldwide are burning millions of hectares of forests and green land affecting the social, ecological, and economical situation. Widely used indices-based threshold methods like Normalized Burned Ratio (NBR) require a huge amount of data preprocessing and are specific to the data capturing source. State-of-the-art deep learning models, on the other hand, are supervised and require domain experts knowledge for labeling the data in huge quantity. These limitations make the existing models difficult to be adaptable to new variations in the data and capturing sources. In this work, we have proposed an unsupervised deep learning based architecture to map the burnt regions of forests by learning features progressively. The model considers small patches of satellite imagery and classifies them into burnt and not burnt. These small patches are concatenated into binary masks to segment out the burnt region of the forests. The proposed system is composed of two modules: 1) a state-of-the-art deep learning architecture for feature extraction and 2) a clustering algorithm for the generation of pseudo labels to train the deep learning architecture. The proposed method is capable of learning the features progressively in an unsupervised fashion from the data with pseudo labels, reducing the exhausting efforts of data labeling that requires expert knowledge. We have used the realtime data of Sentinel-2 for training the model and mapping the burnt regions. The obtained F1-Score of 0.87 demonstrates the effectiveness of the proposed model.

Seagrass ecosystems are pivotal in marine environments, serving as crucial habitats for diverse marine species and contributing significantly to carbon sequestration. Accurate classification of seagrass species from underwater images is imperative for monitoring and preserving these ecosystems. This paper introduces Unsupervised Curriculum Learning (UCL) to seagrass classification using the DeepSeagrass dataset. UCL progressively learns from simpler to more complex examples, enhancing the model's ability to discern seagrass features in a curriculum-driven manner. Experiments employing state-of-the-art deep learning architectures, convolutional neural networks (CNNs), show that UCL achieved overall 90.12 % precision and 89 % recall, which significantly improves classification accuracy and robustness, outperforming some traditional supervised learning approaches like SimCLR, and unsupervised approaches like Zero-shot CLIP. The methodology of UCL involves four main steps: high-dimensional feature extraction, pseudo-label generation through clustering, reliable sample selection, and fine-tuning the model. The iterative UCL framework refines CNN's learning of underwater images, demonstrating superior accuracy, generalization, and adaptability to unseen seagrass and background samples of undersea images. The findings presented in this paper contribute to the advancement of seagrass classification techniques, providing valuable insights into the conservation and management of marine ecosystems. The code and dataset are made publicly available and can be assessed here: https://github.com/nabid69/Unsupervised-Curriculum-Learning—UCL.

This paper presents a Convolutional Neural Networks (CNN) based Unsupervised Curriculum Learning approach for the recognition of water bodies to overcome the stated challenges for remote sensing based RGB imagery. The unsupervised nature of the presented algorithm eliminates the need for labelled training data. The problem is cast as a two class clustering problem (water and non-water), while clustering is done on deep features obtained by a pre-trained CNN. After initial clusters have been identified, representative samples from each cluster are chosen by the unsupervised curriculum learning algorithm for fine-tuning the feature extractor. The stated process is repeated iteratively until convergence. Three datasets have been used to evaluate the approach and show its effectiveness on varying scales: (i) SAT-6 dataset comprising high resolution aircraft images, (ii) Sentinel-2 of EuroSAT, comprising remote sensing images with low resolution, and (iii) PakSAT, a new dataset we created for this study. PakSAT is the first Pakistani Sentinel-2 dataset designed to classify water bodies of Pakistan. Extensive experiments on these datasets demonstrate the progressive learning behaviour of UCL and reported promising results of water classification on all three datasets. The obtained accuracies outperform the supervised methods in domain adaptation, demonstrating the effectiveness of the proposed algorithm.

The worldwide electronics market is focusing on developing innovative technologies that can lead to denser, more resilient, and tighter board-level integration. The consumer electronics market is trending toward miniaturization, with HDI-PCBs dominating. Electronics shrinking and scaling technology is the prime concern of all manufacturers. The PCBA industry is transforming its production practices which can reduce the solder joints, limit the usage of discrete and bulky components, reduce the packaging factor of printed boards by accommodating the maximum number of ICs, minimize the assembly span, optimize the latency, and so on. However, developments in production processes in the PCB manufacturing industry need more attention than those in  Silicon-based (ICs) fabrications. One of the issues in PCB fabrication is utilizing conventional metallization approaches. The majority of manufacturers continue to use standard Copper(Cu) laminates on the base substrate and lithography methods to shape the structures.In recent manufacturing technologies, semi-Additive process (SAP) or modified-SAP (mSAP) methods are being adopted to replace traditional subtractive print-and-etch procedures. To scale down the Lines and Spaces (L\&S) on PCBs comparable to that of IC-level, most smartphone makers use Substrate-like PCB (SLP) using mSAP methods. However, subtractive patterning has been used in the intermediate stages of fabrication in those methods. This thesis demonstrates a fully additive selective metallization-based production approach to bridge this technology gap between IC-level and board-level fabrications. The fabrication process has given the name 'Sequential Build-Up Covalent Bonded Metallisation' (SBU-CBM) method.

This dissertation presents a new approach to Cu metallization using a significant step reducing-pattern-transfer process. The patterning method activates a seed layer of CBM polymer chains on a polymer surface with optimal UV-Laser settings. This surface modification enables a strong Copper (Cu) bonding onto the modified surface by Cu-plating. The suggested approach generated a 2.5D surface pattern using a micrometer via laser ablation and subsequent sub-micrometer laser lithography. Furthermore, the surface characterization of each step involved in the fabrication process is analysed and presented to show the sequential growth of layers on top of each other. To investigate the mechanism of the process at the interfaces, characterizations such as EDS, SEM, and XRD characterizations were performed. This PCB manufacturing method can selectively add metallic layers to the finest feature sizes at considerably lower temperatures. Overall, the thesis has addressed two critical aspects i.e. miniaturization of interconnects at board-level and the feasibility of a fully-additive production approach for electronics packaging.

First, a subtractive method is shown to achieve Copper interconnects with feature size 3.0$\mu$m. This miniaturization corresponds to 70\% reduction in the feature size from 20 $\mu$m to 3 $\mu$m. Next, the proposed additive production process has produced Cu interconnects with feature sizes of 2.5 $\mu$m L\&S and via of diameter 10 $\mu$m. The scaling of the interconnects was achieved by optimizing the process parameters involved in the proposed fabrication recipe.

Second, the sequential build-up (SBU) procedure is adopted to realize the embedded passives with the minimum possible feature size ($<$ 10 $\mu$m). An embedded capacitor and a planar inductor were fabricated. The proposed method can be employed to achieve any desirable pattern on FR-4, and a few of them are shown in the thesis. This additive technique can further be investigated through electrical and reliability assessment to make it an industrially accepted method.

Consumer electronics market is escalating towards the miniaturization and the use of HDI-PCBs is dominating. Thus, the production technologies are adapting the Semi-Additive process (SAP) or modified-SAP (mSAP) methods over conventional subtractive print-and-etch methods. Most of the Smartphone manufacturers are using Substrate-like PCB (SLP) with mSAP techniques to scale down the Lines and Spaces (L&S) on PCBs equivalent to ICs. However, those processes still involve subtractive patterning in the intermediate stages of fabrication. In this paper, a fully additive multi-layer patterning process using an electroless copper plating has been investigated. This patterning process is based on modifying a polymer surface by activating a seed layer of grafting polymer chains on it using optimized UV-Laser parameters. This surface modification enables a strong bonding of Copper (Cu) onto the modified surface by Cu-plating. Using a micrometer via laser ablation and subsequent sub-micrometer laser lithography a 2.5D surface pattern has been achieved with the proposed technique.So far, using the proposed additive production process the feature sizes of 2.5 μm L&S and via of diameter 10 μm have been achieved.The via ablation and pattering were done by using 266nm and 375nm laser sources respectively.The substrates used are standard FR4 material and a layer of polyurethane of thickness 35μm coated on top of it. Analysis of the process parameters and their optimization has been done by factorial design method using Design Expert 12.0 software to show their contribution and significance in the production process.

Advancements in production techniques in PCB manufacturing industries are still required as compared to silicon-ICs fabrications. One of the concerned areas in PCBs fabrication is the use of conventional methodologies for metallization. Most of the manufacturers are still using the traditional Copper (Cu) laminates on the base substrate and patterning the structures using lithography processes. As a result, significant amounts of metallic parts are etched away during any mass production process, causing unnecessary disposables leading to pollution. In this work, a new approach for Cu metallization is demonstrated with considerable step-reducing pattern-transfer mechanism. In the fabrication steps, a seed layer of covalent bonded metallization (CBM) chemistry on top of a dielectric epoxy resin is polymerized using actinic radiation intensity of a 375 nm UV laser source. The proposed method is capable of patterning any desirable geometries using the above-mentioned surface modification followed by metallization. To metallize the patterns, a proprietary electroless bath has been used. The metallic layer grows only on the selective polymer-activated locations and thus is called selective metallization. The highlight of this production technique is its occurrence at a low temperature (20–45 °C). In this paper, FR-4 as a base substrate and polyurethane (PU) as epoxy resin were used to achieve various geometries, useful in electronics packaging. In addition, analysis of the process parameters and some challenges witnessed during the process development are also outlined. As a use case, a planar inductor is fabricated to demonstrate the application of the proposed technique.

Miniaturization is the call of the day. Electronics shrinking and scaling technology is the priority of all manufacturers. PCBA Industry is working towards the elimination of solder joints, reduction in use of discrete and bulky components, lowering of assemble span, minimized latency etc. Embedded passive technology is playing a significant role in this roadmap by providing better signal performance, reduced parasitic and crosstalk. In this work, the primary focus is to develop a cost-efficient and flexible fabrication methodology that will be suitable for bulk production. A sequential build up (SBU) procedure is adopted with an additive lithography process to realize the passives with minimum possible feature size (<; 10 μm). A low cost insulating material, promising grafting solution and Laser assisted writing machine with optimized fabrication parameters are the highlights of this production method. A Computer Aided Design (CAD) software i.e. clewin is used during this process to pattern the mask for the entire process. Covalent bonded metallization (CBM) is the key process for the adhesion of copper layer on the desired site of the pattern. In the CBM process, a polymer surface is modified by grafting. The position of the surface modification is optically defined using a laser lithography system. Such surface modified samples are, then treated in an electroless copper process. Resulting in copper metallization only at the locations with a CBM modified surface. The verification of the copper deposition on the substrate is investigated using a high-resolution microscope followed by scanning electron microscopy (SEM). The confirmation of passive formation has been checked using kethley's source (electrical two-probe measurement). The first-order measured results showed the capacitance formed in the range of 0.3-8 pF. Further concrete measurements using standard methods are undergoing. One of the key advantage of this proposed process is its easiness and feasibility of at room temperature.

As per the latest roadmap of iNEMI, the global electronics market is emphasizing to identify disruptive technologies that can contribute towards denser, robust and tighter integration on the board level. Therefore, reduction in packaging factor of printed board can accommodate greater number of ICs to support miniaturization. This paper has shown an experimental method to pattern the metallic layer on a Printed circuit Board (PCB) to the smallest feature size. To investigate this, a commercially available FR-4 PCB with photosensitive material coat and a Copper (Cu) layer on it, is used. A reverse-mode Laser assisted writing is implemented to pattern the desired copper tracks. Soon after, a well-controlled development and chemical etching of the Laser-activated regions are done using Sodium Hydroxide solution followed by an aqueous solution of Sodium Persulfate. Current PCB interconnects used by the industries are of the order (~20 μm). Whereas the present work is a contribution towards achieving Copper interconnects with feature size 3.0μm. This miniaturization corresponds to 70% reduction in the feature size from 20 μm to 3μm. The natural adhesion of the Cu layer has remained intact even after the etching, shows the efficiency of the method adopted. Also, variation in the parameters such as etching time, etchant solution concentrations, temaperature, gain and exposure time of Laser beam and their corresponding effects are discussed. Other highlights of this subtractive method includes its cost-efficiency, lesser production time and repeatability.

To bridge the technology gap between IC-level and board-level fabrications, a fully additive selective metallization has already been demonstrated in the literature. In this article, the surface characterization of each step involved in the fabrication process is outlined with bulk metallization of the surface. This production technique has used polyurethane as epoxy resin and proprietary grafting chemistry to functionalize the surface with covalent bonds on an FR-4 base substrate. The surface was then metalized using an electroless copper (Cu) bath. This sequential growth of layers on top of each other using an actinic laser beam and palladium (Pd) ions to deposit Cu is analyzed. State-of-the-art material characterization techniques were employed to investigate process mechanism at the interfaces. Density functional theory calculations were performed to validate the experimental evidence of covalent bonding of the layers. This manufacturing approach is capable of adding metallic layers in a selective manner to the printed circuit boards at considerably lower temperatures. A complete analysis of the process using bulk deposition of the materials is illustrated in this work.

We take a step towards addressing the under-representation of the African continent in NLP research by bringing together different stakeholders to create the first large, publicly available, high-quality dataset for named entity recognition (NER) in ten African languages. We detail the characteristics of these languages to help researchers and practitioners better understand the challenges they pose for NER tasks. We analyze our datasets and conduct an extensive empirical evaluation of state-of-the-art methods across both supervised and transfer learning settings. Finally, we release the data, code, and models to inspire future research on African NLP.

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.

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.

This work highlights some important factors for consideration when developing word vector representations and data-driven conversational systems. The neural network methods for creating word embeddings have gained more prominence than their older, count-based counterparts.However, there are still challenges, such as prolonged training time and the need for more data, especially with deep neural networks. Shallow neural networks with lesser depth appear to have the advantage of less complexity, however, they also face challenges, such as sub-optimal combination of hyper-parameters which produce sub-optimal models. This work, therefore, investigates the following research questions: "How importantly do hyper-parameters influence word embeddings’ performance?" and "What factors are important for developing ethical and robust conversational systems?" In answering the questions, various experiments were conducted using different datasets in different studies. The first study investigates, empirically, various hyper-parameter combinations for creating word vectors and their impact on a few natural language processing (NLP) downstream tasks: named entity recognition (NER) and sentiment analysis (SA). The study shows that optimal performance of embeddings for downstream \acrshort{nlp} tasks depends on the task at hand.It also shows that certain combinations give strong performance across the tasks chosen for the study. Furthermore, it shows that reasonably smaller corpora are sufficient or even produce better models in some cases and take less time to train and load. This is important, especially now that environmental considerations play prominent role in ethical research. Subsequent studies build on the findings of the first and explore the hyper-parameter combinations for Swedish and English embeddings for the downstream NER task. The second study presents the new Swedish analogy test set for evaluation of Swedish embeddings. Furthermore, it shows that character n-grams are useful for Swedish, a morphologically rich language. The third study shows that broad coverage of topics in a corpus appears to be important to produce better embeddings and that noise may be helpful in certain instances, though they are generally harmful. Hence, relatively smaller corpus can show better performance than a larger one, as demonstrated in the work with the smaller Swedish Wikipedia corpus against the Swedish Gigaword. The argument is made, in the final study (in answering the second question) from the point of view of the philosophy of science, that the near-elimination of the presence of unwanted bias in training data and the use of foralike the peer-review, conferences, and journals to provide the necessary avenues for criticism and feedback are instrumental for the development of ethical and robust conversational systems.

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.

We reflect on the experiences in organizing and implementing a high-quality Blended Intensive Programme (BIP) as a joint international event. A BIP is a short programme that combines physical mobility with a virtual part. The 6-day event, titled “DigiHealth-AI: Practice, Research, Ethics, and Regulation”, was organized in collaboration with partners from five European nations and support from the EU’s ERASMUS+ programme in November 2023. We introduced a new learning method called ProCoT, involving large language models (LLMs), for preventing cheating by students in writing. We designed an online survey of key questions, which was conducted at the beginning and the end of the BIP. The highlights of the survey are as follows: By the end of the BIP, 84% of the respondents agreed that the intended learning outcomes (ILOs) were fulfilled, 100% strongly agreed that artificial intelligence (AI) benefits the healthcare sector, 62% disagree that they are concerned about AI potentially eliminating jobs in the healthcare sector (compared to 57% initially), 60% were concerned about their privacy when using AI, and 56% could identify, at least, two known sources of bias in AI systems (compared to only 43% prior to the BIP). A total of 541 votes were cast by 40 students, who were the respondents. The minimum and maximum numbers of students who answered any particular survey question at a given period are 25 and 40, respectively.

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, we show that embeddings from relatively smaller corpora sometimes outperform thosefrom larger corpora and we introduce a new Swedish analogy test set and make it publicly available.To achieve good performance in Natural Language Processing (NLP) downstream tasks, several factorsplay important roles: dataset size, the right hyper-parameters, and well-trained embeddings. We utilizethe fastText tool for our experiments. We evaluate both the Swedish and English embeddings that wecreated using intrinsic evaluation (including analogy & Spearman correlation) and compare them with2 common, publicly available embeddings. Our English continuous Bag-of-Words (CBoW)-negativesampling embedding shows better performance compared to the publicly available GoogleNews version.We also describe the relationship between NLP and cognitive science. We contribute the embeddings forresearch or other useful purposes by publicly releasing them.

In this paper, our main contributions are that embeddings from relatively smaller corpora can outperform ones from far larger corpora and we present the new Swedish analogy test set. To achieve a good network performance in natural language processing (NLP) downstream tasks, several factors play important roles: dataset size, the right hyper-parameters, and well-trained embeddings. We show that, with the right set of hyper-parameters, good network performance can be reached even on smaller datasets. We evaluate the embeddings at the intrinsic level and extrinsic level, by deploying them on the Transformer in named entity recognition (NER) task and conduct significance tests. This is done for both Swedish and English. We obtain better performance in both languages on the downstream task with far smaller training data, compared to recently released, common crawl versions; and character n-grams appear useful for Swedish, a morphologically rich language.

In this study, we demonstrate that an open-domain conversational system trained on idioms or figurative language generates more fitting responses to prompts containing idioms. Idioms are a part of everyday speech in many languages and across many cultures, but they pose a great challenge for many natural language processing (NLP) systems that involve tasks such as information retrieval (IR), machine translation (MT), and conversational artificial intelligence (AI). We utilized the Potential Idiomatic Expression (PIE)-English idiom corpus for the two tasks that we investigated: classification and conversation generation. We achieved a state-of-the-art (SoTA) result of a 98% macro F1 score on the classification task by using the SoTA T5 model. We experimented with three instances of the SoTA dialogue model—the Dialogue Generative Pre-trained Transformer (DialoGPT)—for conversation generation. Their performances were evaluated by using the automatic metric, perplexity, and a human evaluation. The results showed that the model trained on the idiom corpus generated more fitting responses to prompts containing idioms 71.9% of the time in comparison with a similar model that was not trained on the idiom corpus. We have contributed the model checkpoint/demo/code to the HuggingFace hub for public access.

Word2Vec is a prominent model for natural language processing tasks. Similar inspiration is found in distributed embeddings (word-vectors) in recent state-of-the-art deep neural networks. However, wrong combination of hyperparameters can produce embeddings with poor quality. The objective of this work is to empirically show that Word2Vec optimal combination of hyper-parameters exists and evaluate various combinations. We compare them with the publicly released, original Word2Vec embedding. Both intrinsic and extrinsic (downstream) evaluations are carried out, including named entity recognition and sentiment analysis. Our main contributions include showing that the best model is usually task-specific, high analogy scores do not necessarily correlate positively with F1 scores, and performance is not dependent on data size alone. If ethical considerations to save time, energy, and the environment are made, then relatively smaller corpora may do just as well or even better in some cases. Increasing the dimension size of embeddings after a point leads to poor quality or performance. In addition, using a relatively small corpus, we obtain better WordSim scores, corresponding Spearman correlation, and better downstream performances (with significance tests) compared to the original model, which is trained on a 100 billion-word corpus.

Word2Vec is a prominent model for natural language processing (NLP) tasks. Similar nspiration is found in distributed embeddings for new state-of-the-art (SotA) deep neural networks.  However, wrong combination of hyper-parameters can produce poor quality vectors. The objective of this work is to empirically show optimal combination of hyper-parameters exists and evaluate various combinations. We compare them with the released, pre-trained original word2vec model. Both intrinsic and extrinsic (downstream) evaluations, including named entity recognition (NER) and sentiment analysis (SA) were carried out. The downstream tasks reveal that the best model is usually task-specific, high analogy scores don’t necessarily correlate positively with F1 scores and the same applies to focus on data alone. Increasing vector dimension size after a point leads to poor quality or performance. If ethical considerations to save time, energy and the environment are made, then reasonably smaller corpora may do just as well or even better in some cases. Besides, using a small corpus, we obtain better human-assigned WordSim scores, corresponding Spearman correlation and better downstream performances (with significance tests) compared to the original model, trained on 100 billion-word corpus.

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.

This paper describes the system used by the Machine Learning Group of LTU in subtask 1 of the SemEval-2022 Task 4: Patronizing and Condescending Language (PCL) Detection. Our system consists of finetuning a pretrained text-to-text transfer transformer (T5) and innovatively reducing its out-of-class predictions. The main contributions of this paper are 1) the description of the implementation details of the T5 model we used, 2) analysis of the successes & struggles of the model in this task, and 3) ablation studies beyond the official submission to ascertain the relative importance of data split. Our model achieves an F1 score of 0.5452 on the official test set.

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.

We investigate five English NLP benchmark datasets (on the superGLUE leaderboard) and two Swedish datasets for bias, along multiple axes. The datasets are the following: Boolean Question (Boolq), CommitmentBank (CB), Winograd Schema Challenge (WSC), Winogender diagnostic (AXg), Recognising Textual Entailment (RTE), Swedish CB, and SWEDN. Bias can be harmful and it is known to be common in data, which ML models learn from. In order to mitigate bias in data, it is crucial to be able to estimate it objectively. We use bipol, a novel multi-axes bias metric with explainability, to estimate and explain how much bias exists in these datasets. Multilingual, multi-axes bias evaluation is not very common. Hence, we also contribute a new, large Swedish bias-labeled dataset (of 2 million samples), translated from the English version and train the SotA mT5 model on it. In addition, we contribute new multi-axes lexica for bias detection in Swedish. We make the codes, model, and new dataset publicly available.

We present a fairly large, Potential Idiomatic Expression (PIE) dataset for Natural Language Processing (NLP) in English. The challenges with NLP systems with regards to tasks such as Machine Translation (MT), word sense disambiguation (WSD) and information retrieval make it imperative to have a labelled idioms dataset with classes such as it is in this work. To the best of the authors’ knowledge, this is the first idioms corpus with classes of idioms beyond the literal and the general idioms classification. Inparticular, the following classes are labelled in the dataset: metaphor, simile, euphemism, parallelism, personification, oxymoron, paradox, hyperbole, irony and literal. We obtain an overall inter-annotator agreement (IAA) score, between two independent annotators, of 88.89%. Many past efforts have been limited in the corpus size and classes of samples but this dataset contains over 20,100 samples with almost 1,200 cases of idioms (with their meanings) from 10 classes (or senses). The corpus may also be extended by researchers to meet specific needs. The corpus has part of speech (PoS) tagging from the NLTK library. Classification experiments performed on the corpus to obtain a baseline and comparison among three common models, including the state-of-the-art (SoTA) BERT model, give good results. We also make publicly available the corpus and the relevant codes for working with it for NLP tasks.

A collaborative research project between the University of Leeds and Luleå University of Technology, Sweden, has taken place aiming to characterize the man-made noise in urban and indoor environments in the Galileo allocated frequency bands: E5 (1191.795 ± 25.575 MHz), E6 (1278.75 ± 20 MHz), and L1 (1575.42 ± 16 MHz), obviously, also covering the GPS L1 and L5 bands. This project has been co-funded by the European GNSS Supervisory Authority (EGSA), with funding from the 6th Framework Program of the European Community for research and technological development. The project includes the development of two receiver systems: the first instrument is based on the use of a spectrum analyzer (SA-based instrument), a wideband GNSS antenna, and a front-end capable of capturing each Galileo band separately using appropriate filtering and switches. The second instrument addresses the issues of cost and portability, providing interference detection and alarm triggering without the need for complex instrument. This is accomplished using low cost components in a small form factor where the instrument is based on a core GNSS front-end. This instrument will only cover the L1/E1 band. The interference measurement is obtained by combining the information from the automatic gain control (AGC) voltage that controls the AGC amplifier gain and the spectrum analysis of the analogue-to-digital converter (ADC) output raw data. The AGC information will be very important for detecting the presence of wideband interference signals where this will be difficult using spectrum analysis (in contrast to the case to narrowband interference signals). Control and data logging from both instruments are performed using a laptop computer where the spectrum analyzer traces and the FE-based instrument data are recorded for offline analysis via a suite of MATLAB® scripts. This paper describes the spectrum survey conducted at various indoor and urban locations, operationally significant to GNSS, in the North of the UK. The survey sites were selected to obtain geographically diverse measurement results and provide a general representation of the spectral environment. In addition, the temporal variation of man-made noise (MMN) is considered, this latter being correlated with the human activity at the measurement site, by performing the measurements day and night, weekdays and weekends.

This article describes analytical work carried out in a pilot project for the Swedish Space Data Lab (SSDL), which focused on monitoring drought in the Mälardalen region in central Sweden. Normalized Difference Vegetation Index (NDVI) and the Moisture Stress Index (MSI) – commonly used to analyse drought – are estimated from Sentinel 2 satellite data and averaged over a selection of seven grassland areas of interest. To derive a complete time-series over a season that interpolates over days with missing data, we use Gaussian Process Regression, a technique from multivariate Bayesian analysis. The analysis show significant differences at 95% confidence for five out of seven areas when comparing the peak drought period in the dry year 2018 compared to the corresponding period in 2019. A cross-validation analysis indicates that the model parameter estimates are robust for temporal covariance structure (while inconclusive for the spatial dimensions). There were no signs of over-fitting when comparing in-sample and out-of-sample RMSE.

This paper presents a deep learning benchmark on a complex dataset known as KFUPM Handwritten Arabic TexT (KHATT). The KHATT data-set consists of complex patterns of handwritten Arabic text-lines. This paper contributes mainly in three aspects i.e., (1) pre-processing, (2) deep learning based approach, and (3) data-augmentation. The pre-processing step includes pruning of white extra spaces plus de-skewing the skewed text-lines. We deploy a deep learning approach based on Multi-Dimensional Long Short-Term Memory (MDLSTM) networks and Connectionist Temporal Classification (CTC). The MDLSTM has the advantage of scanning the Arabic text-lines in all directions (horizontal and vertical) to cover dots, diacritics, strokes and fine inflammation. The data-augmentation with a deep learning approach proves to achieve better and promising improvement in results by gaining 80.02% Character Recognition (CR) over 75.08% as baseline.

Recent progress in deep learning has led to accurate and efficient generic object detection networks. Training of highly reliable models depends on large datasets with highly textured and rich images. However, in real-world scenarios, the performance of the generic object detection system decreases when (i) occlusions hide the objects, (ii) objects are present in low-light images, or (iii) they are merged with background information. In this paper, we refer to all these situations as challenging environments. With the recent rapid development in generic object detection algorithms, notable progress has been observed in the field of deep learning-based object detection in challenging environments. However, there is no consolidated reference to cover the state of the art in this domain. To the best of our knowledge, this paper presents the first comprehensive overview, covering recent approaches that have tackled the problem of object detection in challenging environments. Furthermore, we present a quantitative and qualitative performance analysis of these approaches and discuss the currently available challenging datasets. Moreover, this paper investigates the performance of current state-of-the-art generic object detection algorithms by benchmarking results on the three well-known challenging datasets. Finally, we highlight several current shortcomings and outline future directions.

I publikationen har vi undersökt användningen av sensorer i en lagerringsslipmaskin för felklassificering och tillståndsövervakning. Föreslagen metod kombinerar domänkunskap om processövervakning och tillståndsövervakning för att framgångsrikt uppnå fellägesförutsägelse med hög noggrannhet med endast ett fåtal nyckelsensorer. Denna forskning visar att tillverkningsutrustning kan dra fördel av avancerad databehandling och maskininlärningsteknik.

Slipmaskinen är av typ SGB55 från Lidköping Machine Tools och används i detta fall för att slipa löpbanor på lagerinnerringar av typ SKF-6210 spårkullager. Sensorer för vibration, akustisk emission, kraft och temperatur är installerade för att övervaka maskinens tillstånd under slipning och olika driftsförhållanden. Data insamlas från sensorerna samt maskinens numeriska styrenhet under drift. Utvalda producerade kvalitetsparametrar mäts efter slipoperationen.

Technical failures in machines are major sources of unplanned downtime in any production and result in reduced efficiency and system reliability. Despite the well-established potential of Machine Learning techniques in condition-based maintenance (CBM), the lack of access to failure data in production machines has limited the development of a holistic approach to address machine-level CBM. This paper presents a practical approach for failure mode prediction using multiple sensors installed in a bearing ring grinder for process control as well as condition monitoring. Bearing rings are produced in a set of 7 experimental runs, including 5 frequently occurring production failures in the critical subsystems. An advanced data acquisition setup, implemented for CBM in the grinder, is used to capture information about each individual grinding cycle. The dataset is pre-processed and segmented into grinding cycle stages before time and frequency domain feature extraction. A sensor ranking algorithm is proposed to optimize feature selection for failure classification and the installation cost. Random forest models, benchmarked as best performing classifiers, are trained in a two-step classification framework. The presence of failure mode is predicted in the first step and the failure mode type is identified in the second step using the same feature set. Defining the feature set in the failure detection step improves the predictor generalization with the classifiers’ performance accuracy of 99%99% on the test dataset. The presented approach demonstrates an efficient failure mode classification by selecting crucial sensors resulting in a cost-effective CBM implementation in a bearing ring grinder.

Grinding processes’ stochastic nature poses a challenge in predicting the quality of the resulting surfaces. Post-production measurements for form, surface roughness, and circumferential waviness are commonly performed due to infeasibility in measuring all quality parameters during the grinding operation. Therefore, it is challenging to diagnose the root cause of quality deviations in real-time resulting from variations in the machine’s operating condition. This paper introduces a novel approach to predict the overall quality of the individual parts. The grinder is equipped with sensors to implement condition-based maintenance and is induced with five frequently occurring failure conditions for the experimental test runs. The crucial quality parameters are measured for the produced parts. Fuzzy c-means (FCM) and Hotelling’s T-squared (T2) have been evaluated to generate quality labels from the multi-variate quality data. Benchmarked random forest regression models are trained using fault diagnosis feature set and quality labels. Quality labels from the T2 statistic of quality parameters are preferred over FCM approach for their repeatability. The model, trained from T2 labels achieves more than 94% accuracy when compared to the measured ring disposition. The predicted overall quality using the sensors’ feature set is compared against the threshold to reach a trustworthy maintenance decision.

This thesis presents the foundations of a method for estimating fibre properties of pulp suitable for online application in the pulp and paper industry. In the pulp and paper industry, increased efficiency and greater paper quality control are two of the industry's main objectives. It is proposed that online fibre property measurements are a means of achieving progress in both of these objectives. Optical based systems that provide valuable geometric data on the fibres and other pulp characteristics are commercially available. However, measurements of the elastic properties of the fibres are not currently implemented using these systems. To fill this gap an ultrasound based system for measuring the elastic properties of the wood fibres in pulp is proposed. Ultrasound propagation depends on the elastic properties of a solid. Hence attenuation measurements from suspensions of fibres depend on their elastic properties. The method is based on solving the inverse problem where the output is known and the objective is to establish the inputs. In this case, attenuation is measured and a model of attenuation based on ultrasound scattering is developed. A search algorithm is used for finding elastic properties that minimize the error between the model and measured attenuation. The results of the search are estimates of the elastic properties of the fibres in suspension. The results show resonance peaks in the attenuation, in the frequency region tested, for fibres with radii of the order of 10 microns. These peaks are found in both the measured and modelled attenuation spectra. Further investigation of these resonances suggests that they are due to modes of vibration in the fibre where the fibre modelled as an infinitely long cylinder. These resonances are shown to aid in the identification of the elastic properties. The attenuation is found to depend heavily on the geometry of the fibres. Hence fibre geometry, which can be obtained from online optical fibre measurement system, provides the key to extracting the elastic properties from the attenuation signal. Studies are also carried out on the effect of viscosity on attenuation as well as the differences in attenuation between hollow and solid synthetic fibres in suspensions. The measurement method is also applied to hardwood and softwood Kraft pulps. The results of these studies show that using the model derived in the thesis and attenuation measurements, estimates of the elastic properties can be obtained. The elastic property estimates for synthetic fibres agree well with values from other methods. The elastic property estimates for pulps require further validation due to the difficulty in comparing between different testing methods and different types of pulp. The conclusions, based on the work so far and under three realisable conditions, are that the shear modulus and the transverse Young's modulus of pulp fibres can be measured. Once these conditions are met a system based on this method can be implemented. By doing this the industry would benefit from the increase in paper quality control and energy saving such system could provide.

For the paper and pulp industry in Sweden and Finland to remain competitive against countries with lower overheads, they have to constantly strive to improve the quality and the efficiency of the manufacturing processes. One of the ways of doing this is to introduce sensors that will provide valuable online feedback on the characteristics of the pulp so that adjustments can be made to optimise the manufacturing process. The measurement method proposed in this thesis is based on ultrasound, since it is rapid, inexpensive, non-destructive and non-intrusive. Thus could be done online. Since ultrasound propagation and attenuation depends on the material properties through which is propagates, it has the potential to provide measurements of material properties such as pulp fibre density and elasticity. The aim of this thesis is to investigate the possibility of using ultrasound to measure pulp fibre material properties. The idea is to solve the inverse problem of estimating these properties from attenuation measurements and to establish the degree of accuracy to which this can be done. Firstly a model is developed and is tested with synthetic fibres to establish is validity. It is then used to solve the inverse problem of estimating material properties from attenuation measurements, again with synthetic fibres, to test the accuracy to which these properties can be estimated. Resonance peaks in the frequency response of the attenuation were found. On closer investigation it was established that the location of these peaks in the frequency domain is sensitive to the diameter of the fibres and their material properties. If the diameter is known, these peaks improve the accuracy of the estimation process. The results of the estimation process for synthetic fibre suspensions show values for the shear modulus are within known ranges but the estimation of Poisson's ratio and Young's modulus is poor. Improving the model or the estimation procedure may lead to better results. For the method as it is to have application in the paper and pulp industry there are certain conditions that need to be fulfilled. These are that we find peaks in the frequency response of the attenuation in pulp, know the diameter distribution of the fibres and the hollow nature of the fibres does not significantly alter the results. We can then, potentially, be able to establish the shear modulus of the pulp fibres. If the shear modulus is a factor in paper quality, we may be close to an online measurement of paper pulp quality using ultrasonic spectroscopy. Improving the model may allow us to estimate further properties and take into account the fibres being hollow. The thesis consists of two parts. The first part includes an overview of the pulp and paper industry and current testing methods, background theory on which the model is based and an overview of the model that is used in predicting ultrasound attenuation. There then follows a summary of the work done, some addition points are raised in the discussion before drawing conclusions. Finally we discuss what needs to be done to take this further. The second part contains a collection of four papers describing the research.

Here light scattering theory is used to assess the size distribution in a suspension of cellulose as it is fibrillated from micro-scaled to nano-scaled fibres. A model based on Monte carlo simulations of the scattering of photons by different sizes of cellulose fibres was used to predict the UV-IF spectrum of the suspensions. Bleached cellulose hardwood pulp was tested and compared to the visually transparent tempo-oxidised hardwood cellulose nanofibres (CNF) suspension. The theoretical results show that different diameter size classes exhibit very different scattering patterns. These classes could be identified in the experimental results and used to establish the size class dominating the suspension. A comparison to AFM/microscope size distribution was made and the results indicated that using the UV-IF light scattering spectrum maybe more reliable that size distribution measurement using AFM and microscopy on dried CNF samples. The UV-IF spectrum measurement combined with the theoretical prediction can be used even at this initial stage of development of this model to assess the degree of fibrillation when processing CNF.

An analytical model for use in the inverse problem of estimating material properties of suspended fibres from ultrasonic attenuation has been developed. The ultrasound attenuation is derived theoretically from the energy losses arising when a plane wave is scattered and absorbed off an infinitely long, isotropic, viscoelastic cylinder. By neglecting thermal considerations and assuming low viscosity in the suspending fluid, we can make additional assumptions that provide us with a tractable set of equations that can be solved analytically. The model can then be to used in inverse methods of estimating material properties. We verify the model with experimentally obtained values of attenuation for saturated Nylon fibres. The experimental results from Nylon fibres show local peaks in the attenuation which are thought to be due to the resonant absorption at the eigenfrequencies of the fibres. The results of the experiments show that the model is sufficiently sensitive to detect differences in different types of Nylon. Applications for suspended fibre characterization can be found in the paper manufacturing industry.