Transformer-based language models have been changing the modern Natural Language Processing (NLP) landscape for high-resource languages such as English, Chinese, Russian, etc. However, this technology does not yet exist for any Ghanaian language. In this paper, we introduce the first of such models for Twi or Akan, the most widely spoken Ghanaian language. The specific contribution of this research work is the development of several pretrained transformer language models for the Akuapem and Asante dialects of Twi, paving the way for advances in application areas such as Named Entity Recognition (NER), Neural Machine Translation (NMT), Sentiment Analysis (SA) and Part-of-Speech (POS) tagging. Specifically, we introduce four different flavours of ABENA -- A BERT model Now in Akan that is fine-tuned on a set of Akan corpora, and BAKO - BERT with Akan Knowledge only, which is trained from scratch. We open-source the model through the Hugging Face model hub and demonstrate its use via a simple sentiment classification example.

Arm, the leader in chips used in everything from mobile devices to supercomputers, has unveiled Armv9, the company's first major architectural change in a decade. The new designs should result in 30% faster performance over the next two chip generations. Arm is a chip architecture company that licenses its designs to others, and its customers have shipped more than 100 billion chips in the past five years. Nvidia is in the midst of acquiring Cambridge, United Kingdom-based Arm for $40 billion, but the deal is waiting on regulatory approvals. In a press briefing, Arm CEO Simon Segars said Armv9 will be the base for the next 300 billion Arm-based chips.

As we rely on machine learning (ML) models to make more consequential decisions, the issue of ML models perpetuating or even exacerbating undesirable historical biases (e.g. In this paper, we focus on the problem of detecting violations of individual fairness in ML models. We formalize the problem as measuring the susceptibility of ML models against a form of adversarial attack and develop a suite of inference tools for the adversarial cost function. The tools allow auditors to assess the individual fairness of ML models in a statistically-principled way: form confidence intervals for the worst-case performance differential between similar individuals and test hypotheses of model fairness with (asymptotic) non-coverage/Type I error rate control. The problem of bias in machine learning systems is at the forefront of contemporary ML research. Numerous media outlets have scrutinized machine learning systems deployed in practice for violations of basic societal equality principles (Angwin et al., 2016; Dastin, 2018; Vigdor, 2019). In response researchers developed many formal definitions of algorithmic fairness along with algorithms for enforcing these definitions in ML models (Dwork et al., 2011; Hardt et al., 2016; Berk et al., 2017; Kusner et al., 2018; Ritov et al., 2017; Yurochkin et al., 2020). Despite the flurry of ML fairness research, the basic question of assessing fairness of a given ML model in a statistically principled way remains largely unexplored. In this paper we propose a statistically principled approach to assessing individual fairness (Dwork et al., 2011) of ML models.

With there being many technical and ethical issues with artificial intelligence (AI) that involve marginalized communities, there is a growing interest for design methods used with marginalized people that may be transferable to the design of AI technologies. Participatory design (PD) is a design method that is often used with marginalized communities for the design of social development, policy, IT and other matters and solutions. However, there are issues with the current PD, raising concerns when it is applied to the design of technologies, including AI technologies. This paper argues for the use of PD for the design of AI technologies, and introduces and proposes a new PD, which we call agile participatory design, that not only can could be used for the design of AI and data-driven technologies, but also overcomes issues surrounding current PD and its use in the design of such technologies.

We consider the task of enforcing individual fairness in gradient boosting. Gradient boosting is a popular method for machine learning from tabular data, which arise often in applications where algorithmic fairness is a concern. At a high level, our approach is a functional gradient descent on a (distributionally) robust loss function that encodes our intuition of algorithmic fairness for the ML task at hand. Unlike prior approaches to individual fairness that only work with smooth ML models, our approach also works with non-smooth models such as decision trees. We show that our algorithm converges globally and generalizes. We also demonstrate the efficacy of our algorithm on three ML problems susceptible to algorithmic bias.

These subprocesses are interrelated, with mapping considered to be the pivotal process (Gentner, 1983). Mapping may play a role in retrieval, as mapping a target analog to multiple potential source analogs stored in memory can help identify one or more that seems promising; and the correspondences computed by mapping support subsequent inference and schema induction. Thus, because of its centrality to analogical reasoning, the present paper focuses on the process of mapping between two analogs. We also consider the possible role that mapping may play in analog retrieval. Computational Approaches to Analogy Computational models of analogy have been developed in both artificial intelligence (AI) and cognitive science over more than half a century (for a recent review and critical analysis, see Mitchell, 2021). These models differ in many ways, both in terms of basic assumptions about the constraints that define a "good" analogy for humans, and in the detailed algorithms that accomplish analogical reasoning. For our present purposes, two broad approaches can be distinguished. The first approach, which can be termed representation matching, combines mental representations of structured knowledge about each analog with a matching process that computes some form of relational similarity, yielding a set of correspondences between the elements of the two analogs. The structured knowledge about an analog is typically assumed to approximate the content of propositions expressed in predicate calculus; e.g., the instantiated relation "hammer hits nail" might be coded as hit (hammer, nail).

Recently, there are unprecedented data growth originating from different online platforms which contribute to big data in terms of volume, velocity, variety and veracity (4Vs). Given this nature of big data which is unstructured, performing analytics to extract meaningful information is currently a great challenge to big data analytics. Collecting and analyzing unstructured textual data allows decision makers to study the escalation of comments/posts on our social media platforms. Hence, there is need for automatic big data analysis to overcome the noise and the non-reliability of these unstructured dataset from the digital media platforms. However, current machine learning algorithms used are performance driven focusing on the classification/prediction accuracy based on known properties learned from the training samples. With the learning task in a large dataset, most machine learning models are known to require high computational cost which eventually leads to computational complexity. In this work, two supervised machine learning algorithms are combined with text mining techniques to produce a hybrid model which consists of Na\"ive Bayes and support vector machines (SVM). This is to increase the efficiency and accuracy of the results obtained and also to reduce the computational cost and complexity. The system also provides an open platform where a group of persons with a common interest can share their comments/messages and these comments classified automatically as legal or illegal. This improves the quality of conversation among users. The hybrid model was developed using WEKA tools and Java programming language. The result shows that the hybrid model gave 96.76% accuracy as against the 61.45% and 69.21% of the Na\"ive Bayes and SVM models respectively.

For unsupervised image-to-image translation, we propose a discriminator architecture which focuses on the statistical features instead of individual patches. The network is stabilized by distribution matching of key statistical features at multiple scales. Unlike the existing methods which impose more and more constraints on the generator, our method facilitates the shape deformation and enhances the fine details with a greatly simplified framework. We show that the proposed method outperforms the existing state-of-the-art models in various challenging applications including selfie-to-anime, male-to-female and glasses removal. The code will be made publicly available.

Among the essential body functions like breathing, eating or drinking, sleeping is probably the most problematic one nowadays. According to the US government through its Centers for Control of Disease and Prevention (CDC), about 9 million citizens have frequent problems to develop good quality sleep and end up resorting to sleeping pills (Ford et al. 2014). In parallel, recent studies (Stranges et al. 2012; Chong et al. 2013) have estimated that at least 15% of adult population might have some kind of sleeping problem or poor-quality sleep as a result of a number of issues. Moreover, the World Health Organization (WHO) (2015) claimed that a good quality sleep was one of the most important factors for good health while sleeping problems were directly related to other diseases, including depression, stress or early cardiac diseases. As a consequence, new units focused on the study and treatment of sleeping problems have been created in hospitals all over the world. The physicians in these units have as their main tool for their work the records obtained during their patients' sleep. These records, called polysomnography (PSG), may include a great variety of signals such as Electrocardiograms, Electroencephalograms, respiratory signals or movement records. Among these signals, the most important one is the Electroencephalogram (EEG) because it is the most reliable to determine the sleep stage a patient is in. The interpretation of an EEG is a highly time-consuming activity (Akben and Alkan 2016), which usually requires a specialist and it is deeply dependent on the expert's expertise.

Artificial intelligence (AI) technologies have dramatically advanced in recent years, resulting in revolutionary changes in people's lives. Empowered by edge computing, AI workloads are migrating from centralized cloud architectures to distributed edge systems, introducing a new paradigm called edge AI. While edge AI has the promise of bringing significant increases in autonomy and intelligence into everyday lives through common edge devices, it also raises new challenges, especially for the development of its algorithms and the deployment of its services, which call for novel design methodologies catered to these unique challenges. In this paper, we provide a comprehensive survey of the latest enabling design methodologies that span the entire edge AI development stack. We suggest that the key methodologies for effective edge AI development are single-layer specialization and cross-layer co-design. We discuss representative methodologies in each category in detail, including on-device training methods, specialized software design, dedicated hardware design, benchmarking and design automation, software/hardware co-design, software/compiler co-design, and compiler/hardware co-design. Moreover, we attempt to reveal hidden cross-layer design opportunities that can further boost the solution quality of future edge AI and provide insights into future directions and emerging areas that require increased research focus.