Browsing by Subject "Artificial Intelligence"

Sort by: Order: Results:

Now showing items 1-17 of 17
  • Flores, Huber; Nurmi, Petteri; Hui, Pan (IEEE, 2019)
    International Conference on Pervasive Computing and Communications
    On-Device AI is an emerging paradigm that aims to make devices more intelligent, autonomous and proactive by equipping them with machine and deep learning routines for robust decision making and optimal execution in devices' operations. On-Device intelligence promises the possibility of computing huge amounts of data close to its source, e.g., sensor and multimedia data. By doing so, devices can complement their counterpart cloud services with more sophisticated functionality to provide better applications and services. However, increased computational capabilities of smart devices, wearables and IoT devices along with the emergence of services at the Edge of the network are driving the trend of migrating and distributing computation between devices. Indeed, devices can reduce the burden of executing resource intensive tasks via collaborations in the wild. While several work has shown the benefits of an opportunistic collaboration of a device with others, not much is known regarding how devices can be organized as a group as they move together. In this paper, we contribute by analyzing how dynamic group organization of devices can be utilized to distribute intelligence on the moving Edge. The key insight is that instead of On-Device solutions complementing with cloud, dynamic groups can be formed to complement each other in an On-Multi-Device manner. Thus, we highlight the challenges and opportunities from extending the scope of On-Device AI from an egocentric view to a collaborative, multi-device view.
  • Malmberg, Otso (Helsingin yliopisto, 2022)
    Artificial Intelligence (AI) has become an ubiquitous technology in society with ever more diverse applications. Different AI systems have become especially pervading in the domain of creative arts, where they can near-autonomously generate content. For example, there is a number of software that allow generation of musical output with minimal specifications from the user of the AI system. Where the creative process combines human and machine labor, results of such joint efforts can be called ‘AI-assisted output’. Ascertaining the legal status of AI-assisted output forms the crux of this Master’s thesis. On one hand, the research is concerned with the copyrightability of AI-assisted output. This question pertains to musical output produced with the assistance of illustrative AI systems of AIVA and Jukebox. The research adopts a legal comparative method in this regard: it seeks to assess how the copyright laws of EU and England could accommodate the AI-assisted output. Evidently, the focus in the analyses of the respective systems will revolve around their conceptions of originality, the key requirement to qualify for copyright protection. As a counterpart for the legal comparative study into the feasibility of copyright protection of AI-assisted output, the second prong of the research probes into the desirability of protecting such output. In other words, the research seeks to establish the merits of protecting AI-assisted outputs by juxtaposing the underlying, theoretical rationales of copyright law against AI-assisted output. To this end, the research aims to survey how the relationship between the human author and AI system interacts with the different rationales of copyright. Against this backdrop, the research preludes by looking into the foundations of AI and the selected legal systems. AI is examined retrospectively with a view on how the technology has emerged and how machine learning, the most prominent method of modern AI, has significantly increased the autonomy of AI systems. The illustrative AI systems of the research, AIVA and Jukebox, match the foregoing trends. However, in exploring the features of the systems, variance can be perceived in the amount of influence the user may be able to exert in the creative process of the musical output. The research then proceeds to provide a thorough picture of the framework of copyright law. In the first half of the section, the origins of the law are traced back to the civil law and common law traditions that underlie most copyright systems. Thereafter, the utilitarian and deontological rationales that underpin those systems are discussed including an exposition of the legal and economic discipline’s relationship with copyright law for later discussion. The second half of the section zeroes in on the copyright laws of the selected legal systems. It first notes the influence from the international framework of copyright and some general discussion on AI and its output on that level. Following up, the respective schemes of awarding copyright protection under EU and English law are broken down into their constituent parts. As per alluded, the crux of the analyses are on their respective standards of originality. In the EU, the notion is construed as an exercise of identifying ‘author’s own intellectual creation‘; specifically, the CJEU is concerned with the identification of creative choices that can take place in different phases of the creative process. In England, the standard of originality denotes first and foremost origination: the pertinent artifact should originate from the author and no one else. In addition, the English standard seeks to establish the exercise of skill, labor and judgment in its creation of the right kind. In the analytic section of the research, the illustrative systems of AIVA and Jukebox are then related against the copyright systems. The section begins with an application of the standards of EU and England against the output from the AI systems in abstract; no concrete samples from the systems are studied but rather the focus is on how their users may harness their features. Herewith, the amount of user engagement in the creative process becomes the key in both system; where the user can apply creative choices in the different phases of the production or alternatively exert skill, labor and judgment as opposed to mere generation off output with the systems, there is very likely an original work at stake. Hereby, the two AI systems diverge when examined in isolation; whereas AIVA provides the user with a vast amount of tools to interact with their product, Jukebox has little to offer in terms of post-processing. Accordingly, whereas output from AIVA would very unlikely fail to meet the standards of either jurisdiction, output from Jukebox could potentially be barred where no human intervention can be perceived from the final product. The section concludes with a survey of alternative forms of protections for AI-assisted output. As per the foregoing discussion, it is noted how the deontological rationales have limited application in justifying the protection of output that have minimal human involvement. It follows that the economic-utilitarian grounds are emphasized and options that spur therefrom. The related rights regime, a new sui generis right and alternatives that disregard considerations of originality and authorship are explored with alike conclusions: the alternatives do not serve as a backdoor protection but are rooted in valuing socially beneficial activities. Evidently, with AIVA and Jukebox, it would have to be evidenced that their generated output merits protection from a market failure. In light of the foregoing discussion, there is no such substantiation. Accordingly, it would be best to leave AI-assisted output in the lower end of the spectrum of human interaction well alone to preserve the balance between private and public interests.
  • Leskinen, Markus; Andersson, Sture (2019)
  • Boggia, Michele; Ivanova, Sardana; Linkola, Simo; Toivonen, Hannu; Kantosalo, Anna (The Association for Computational Creativity, 2022)
    We explore the concept of Casual Poetry Creators with the aim of making poetry writing fun and entertaining for the user. We present a simple co-creative interaction design pattern based on constructing poems line by line, suggesting the user a set of line candidates at each step. We also propose objective measures by which a Casual Poetry Creator can evaluate and choose which line candidates to show to the user and sketch out a plan to evaluate the measures and pattern with users.
  • Xiao, Ping; Toivonen, Hannu; Gross, Oskar; Cardoso, Amilcar; Correia, João; Machado, Penousal; Martins, Pedro; Goncalo Oliveira, Hugo; Sharma, Rahul; Pinto, Alexandre Miguel; Diaz, Alberto; Francisco, Virginia; Gervás, Pablo; Hervas, Raquel; León, Carlos; Forth, Jamie; Purver, Matthew; Wiggins, Geraint A.; Miljkovic, Dragana; Podpecan, Vid; Pollak, Senja; Kralj, Jan; Znidarsic, Martin; Bohanec, Marko; Lavrač, Nada; Urbancic, Tanja; Van Der Velde, Frank; Battersby, Stuart (2019)
    Computational creativity seeks to understand computational mechanisms that can be characterized as creative. The creation of new concepts is a central challenge for any creative system. In this article, we outline different approaches to computational concept creation and then review conceptual representations relevant to concept creation, and therefore to computational creativity. The conceptual representations are organized in accordance with two important perspectives on the distinctions between them. One distinction is between symbolic, spatial and connectionist representations. The other is between descriptive and procedural representations. Additionally, conceptual representations used in particular creative domains, such as language, music, image and emotion, are reviewed separately. For every representation reviewed, we cover the inference it affords, the computational means of building it, and its application in concept creation.
  • Tulilaulu, Aurora; Nelimarkka, Matti; Paalasmaa, Joonas; Johnson, Daniel; Ventura, Dan; Myllys, Petri; Toivonen, Hannu (2018)
    Data musicalization is the process of automatically composing music based on given data, as an approach to perceptualizing information artistically. The aim of data musicalization is to evoke subjective experiences in relation to the information, rather than merely to convey unemotional information objectively. This paper is written as a tutorial for readers interested in data musicalization. We start by providing a systematic characterization of musicalization approaches, based on their inputs, methods and outputs. We then illustrate data musicalization techniques with examples from several applications: one that perceptualizes physical sleep data as music, several that artistically compose music inspired by the sleep data, one that musicalizes on-line chat conversations to provide a perceptualization of liveliness of a discussion, and one that uses musicalization in a game-like mobile application that allows its users to produce music. We additionally provide a number of electronic samples of music produced by the different musicalization applications.
  • Luzan, Tetiana (Helsingin yliopisto, 2018)
    Due to an impressive evolution of the AI technologies within the last few decades it has become an integral part of everyday life called for improving and facilitating it. Yet, as a result of this evolutionary process AI’s activity nowadays contains features which require legal regulation in the course of its application. Such request was recognized by the EU legislator. In 2017, the EP voiced a possibility of introducing of a legal status of electronic person for the sufficiently sophisticated robots. Specialists in this sphere, however, gave a hostile reception to such an initiative claiming that legal personhood of AI cannot be fit into the current legal paradigm. The latest EU initiatives pertaining to the legal regulation of AI application reveal that the electronic person is not on the EU’s agenda anymore. This work is dedicated to a negation of the above claim of AI specialists by demonstrating that there are no unsurmountable obstacles for conferring legal personhood on the sufficiently sophisticated AI even though it should not be considered as a person from the philosophical standpoint. Having accepted a possibility of according legal personhood to AI, the next question to answer is should it be recognized as a natural or artificial person (the existing types) or should it be ascribed a legal status of the electronic person. The answer is found in the analysis of the determining characteristics of sufficiently sophisticated AI and their comparison with the features of humans and corporations, bearers of natural and artificial personhood, respectively. The ascription of legal personhood to AI is not aimed at recognition of exceptional qualities of AI per se. It is, instead, called for resolving the existing legal problems of AI application in the business sphere, namely, in contract and tort law, and intellectual property rights. By the way of the conferral of legal personhood on AI it is possible to properly allocate responsibility or attribute authorship. In such a manner, establishment of legal status of the electronic person is discussed as an umbrella solution for various domains of business law that may establish legal certainty and ensure the EU against legal fragmentation.
  • Boggia, Michele; Ivanova, Sardana; Linkola, Simo; Kantosalo, Anna; Toivonen, Hannu (The Association for Computational Creativity, 2022)
    We present methods that produce poetry one line at a time, in a manner that allows simple interaction in human-computer co-creative poetry writing. The methods are based on fine-tuning sequence-to-sequence neural models, in our case mBART. We also consider several internal evaluation measures by which an interactive system can assess and filter the lines it suggests to the user. These measures concern the coherence, tautology, and diversity of the candidate lines. We empirically validate two of them and apply three on the mBART-based poetry generation methods. The results suggest that fine-tuning a pre- trained sequence-tosequence model is a feasible approach, and that the internal evaluation measures help select suitable models as well as suitable lines.
  • Bhardwaj, Shivam (Helsingin yliopisto, 2020)
    The banking and financial sector has often been synonymous with established names, with some having centuries old presence. In the recent past these incumbents have been experiencing a consequential disruption by new entrants and rapidly changing consumer demands. These disruptions to the status quo have been characterised by a shift towards adoption of technology and artificial intelligence particularly in the service and products offered to the end customers. The changing business climate in the financial sector has risen many convoluted questions for the regulators. These complications cover a vast set of issues – from the concerns relating to the privacy of data of the end users to the increasing vulnerability of the financial market, to unproportionally increased compliance requirements for new entrants, all form part of the mesh of questions that have arisen in the wake of new services and operations being designed with the aid and assistance of artificial intelligence, machine learning and big data analytics. It is in this background that this Thesis seeks to explore the trajectory of the development of the legal landscape for regulating artificial intelligence – both in general and specifically in the financial and banking sector, particularly in the European Union. During the analysis, existing legal enactments, such as the General Data Protection Regulation, have been scrutinised and certain observations have been made regarding the areas that still remain unregulated or open to debate under the laws as it stands today. In the same vein, an attempt has been made to explore the emerging discussion on a dedicated legal regime for artificial intelligence in the European Union, and those observations have been viewed from the perspective of the financial sector, thereby creating thematic underpinnings that ought to form part of any legal instrument aiming to optimally regulate technology in the financial sector. To concretise the actual application of such a legal instrument, a European Union member state has been identified and the evolution of the regulatory regime in the financial sector has been discussed with the said member states’ financial supervisory authority, thus highlighting the crucial role of the law making and enactment bodies in creating and sustaining a technologically innovative financial and banking sector. The themes recognised in this Thesis could be the building blocks upon which the future legal discourse on artificial intelligence and the financial sector could be structured.
  • Calderón, Aparna (Helsingin yliopisto, 2020)
    Tiivistelmä – Referat – Abstract Post the financial crisis of 2008, European Union has introduced a plethora of laws to reform the financial system and make it further resilient. While the crisis led to financial reforms that have created heavy load of compliance, it also created a field for innovation led financial services called Fintech. In the post-Covid-19 era, the need for financial institutions and supervisors to speedily and efficiently deal with compliance has become more pronounced, as they brace for the impact of pandemic and focus on other critical tasks. The combination of regulatory compliance load on one side and innovation on other side have made the role of AI critical to regulatory compliance and supervision. On this premise, the thesis discusses existing role of AI and the challenges in successful deployment of AI where it can be scaled to exploit all its abilities. The challenges in deployment of AI are two fold – one that relate to its role in compliance with financial governance framework including AML, PSD2, MiFID II and GDPR etc. requirements, and other that deals with the role of AI within compliance procedures such as reporting, following up with regulatory requirements and policies etc. The discussion aims to identify the gaps in technology such as black box problem and inductive bias in AI, as well as regulatory framework that hamper the deployment and exploitation of AI in compliance and supervision of financial institutions. At the same time, innovations such as RegTech and SupTech enable AI to provide fast solutions to banks, Fintech and supervisory authorities. The technological and regulatory challenges of AI are identified by doing empirical research and applying legal dogmatics, as well as considering socio-economic factors affecting the financial industry. The discussion also notes the vision of a Digital Europe and how the recently announced policy discussions such as the AI strategy, Data Strategy, Cloud Initiative, Digital Finance Strategy etc make a profound impact on the gaps identified, but also leave exposure to other possibilities such as lengthy process of adoption, implementation and testing of these policies. The discussion concludes that successful adoption of the proposed laws, provisions for financial industry in the AI strategy and new and innovative methods such as data ethnography can together solve the technological and regulatory challenges identified as hinderance to scaling up AI role in regulatory compliance.
  • Vuorio, Alpo; Strandberg, Timo; Kovanen, Petri T. (2020)
  • Mirtti, Tuomas; Lahdenne, Pekka; Pitkänen, Esa (2020)
  • Kärkkäinen, Leo (2020)
  • Toivanen, Jukka M.; Järvisalo, Matti; Alm, Olli; Ventura, Dan; Vainio, Martti; Toivonen, Hannu (2019)
    We study transformational computational creativity in the context of writing songs and describe an implemented system that is able to modify its own goals and operation. With this, we contribute to three aspects of computational creativity and song generation: (1) Application-wise, songs are an interesting and challenging target for creativity, as they require the production of complementary music and lyrics. (2) Technically, we approach the problem of creativity and song generation using constraint programming. We show how constraints can be used declaratively to define a search space of songs so that a standard constraint solver can then be used to generate songs. (3) Conceptually, we describe a concrete architecture for transformational creativity where the creative (song writing) system has some responsibility for setting its own search space and goals. In the proposed architecture, a meta-level control component does this transparently by manipulating the constraints at runtime based on self-reflection of the system. Empirical experiments suggest the system is able to create songs according to its own taste.