[diss] Perustieteiden korkeakoulu / SCI

Permanent URI for this collection


Recent Submissions

Now showing 1 - 20 of 1550
  • Item
    Adaptive OSS: Principles and Design of an Adaptive OSS for 5G Networks
    (Aalto University, 2024) Mfula, Harrison; Nurminen, Jukka K., Prof., University of Helsinki, Finland; Tietotekniikan laitos; Department of Computer Science; Perustieteiden korkeakoulu; School of Science; Ylä-Jääski, Antti, Prof., Department of Computer Science, Aalto University, Finland
    In recent years, the rise and continued popularity of connected applications has resulted in explosive growth in the demand for wireless broadband services of high speed, massive capacity, and ultra-low latency, such as video-on-demand services, Internet of Things, and mission critical applications. 5G technology is designed to provide the required connectivity in these applications. As a consequence of its continued success, seamless connectivity has become synonymous to a human right. Suffice to say, at the moment, due to the vast potential benefits of 5G technology, there is a kind of gold rush driving rapid worldwide deployments of 5G networks that has led to a significant gap in investment, research, and development of suitable operation support system (OSS) solutions for daily operation, monitoring and control of 5G networks. Furthermore, as the number of 5G deployments continue to rise, high data traffic volumes and stakeholder expectation of seamless connectivity from anything to anything has become the norm. In this regard, the need for suitable OSS solutions has become critical. This dissertation fills the identified gap in the following way, first, we design a scalable architecture that enables batch and stream processing of high throughput, high volume, and ultra-low-latency data driven OSS solutions to effectively support existing and 5G OSS use cases. Building on the resulting architecture, we extend existing, and in some cases develop new SON algorithms to meet 5G requirements. Particularly, we develop adaptive algorithms which focus on self-configuration, self-optimization, self-healing, and SON-coordination use cases. Furthermore, we introduce solutions for transitioning from the current mainly proprietary OSS hardware to vendor agnostic cloud-native dynamic infrastructure. Lastly, we make digitization of OSS operations more efficient. Specifically, we develop an artificial intelligence based solution (AIOps) for conducting OSS operations efficiently at cloud scale. Using the findings and proposed solutions in this dissertation, vendors, and service providers can design and implement suitable solutions that meet stringent business and technical requirements of applications running on top of 5G networks and beyond.
  • Item
    Parabolic bounded mean oscillation and Muckenhoupt weights
    (Aalto University, 2024) Myyryläinen, Kim; Kinnunen, Juha, Prof., Aalto University, Department of Accounting, Finland; Matematiikan ja systeemianalyysin laitos; Department of Mathematics and Systems Analysis; Nonlinear PDE research group; Perustieteiden korkeakoulu; School of Science; Kinnunen, Juha, Prof., Aalto University, Department of Accounting, Finland
    This thesis further develops the parabolic theory of functions of bounded mean oscillation (BMO) and Muckenhoupt weights motivated by one-sided maximal functions and a doubly nonlinear parabolic partial differential equation of p-Laplace type. The definition of parabolic BMO consists of two conditions on the mean oscillation of a function, one in the past and the other one in the future with a time lag between the estimates. Various parabolic John–Nirenberg inequalities, which give exponential decay estimates for the oscillation of a function, are shown in the natural geometry of the partial differential equation. We extend and complement the existing theory for the parabolic Muckenhoupt Aq weights and obtain a complete theory for the limiting parabolic Muckenhoupt A1 class including factorization and characterization results. In particular, an uncentered parabolic maximal function with a time lag is applied leading to a more streamlined theory. Weighted norm inequalities are shown for the parabolic maximal function which allows us to establish parabolic versions of the Jones factorization and the Coifman–Rochberg characterization. The other endpoint class of parabolic Muckenhoupt A∞ weights is also discussed and new characterizations are discovered in terms of quantitative absolute continuity with a time lag. Furthermore, this is considered from the perspective of parabolic reverse Hölder inequalities. We obtain several characterizations and self-improving properties for the weights satisfying a parabolic reverse Hölder inequality and study their connection to parabolic Muckenhoupt weights. Parabolic Muckenhoupt weights satisfy the parabolic reverse Hölder inequality, whereas the reverse direction is investigated in terms of a parabolic doubling condition with a time lag. Essential tools in the parabolic theory include delicate parabolic Calderón–Zygmund decompositions, good lambda estimates, covering and chaining arguments. In addition to parabolic BMO, different function spaces of BMO type are studied in the setting of metric measure spaces with a doubling measure. We consider the John–Nirenberg space defined via medians and a weak version of the Gurov–Reshetnyak class. Moreover, we show the corresponding John–Nirenberg inequalities and discuss their consequences. The John–Nirenberg lemma for the median-type John–Nirenberg space gives a polynomial decay estimate for the oscillation of a function. On the other hand, the John–Nirenberg lemma for the weak Gurov–Reshetnyak class provides a specific decay estimate.
  • Item
    Investigations of Ionic Functional Soft Matter
    (Aalto University, 2024) Gustavsson, Lotta; Peng, Bo, Dr., Aalto University, Department of Applied Physics, Finland; Teknillisen fysiikan laitos; Department of Applied Physics; Molecular Materials; Perustieteiden korkeakoulu; School of Science; Ikkala, Olli, Prof., Aalto University, Department of Applied Physics, Finland
    Electrostatic interactions play an important role in functional self-assembled structures of both natural and synthetic origin. Such processes are complex and reflect the importance of balancing competing interactions which is crucial in the development of materials for new technological advancements. This thesis presents studies on electrostatic forces in different environments and how they can be harvested in functional and motile materials. The first part of the thesis presents new liquid-crystalline materials based on ionic surfactants and their complexation. Publication I studied the thermotropic liquid crystallinity of zwitterionic amphiphilic molecules. The observed melting points were high and thus the compounds were plasticized using a low-melting ionic liquid, which led to decreased transition temperatures, ionic liquid-crystalline complex formation, and ion-conduction. In Publication II, the complexation of cellulose nanocrystals (CNCs) by a nonionic-anionic surfactant led to nematic liquid-crystalline phase formation both in organic solvent toluene and in the bulk state. The suppression of the chirality of CNCs is of high technological relevance as it would widen the applicability of CNCs as, e.g., reinforcements and optical polarizers. The electrostatic interactions were accounted as the driving force for the material properties and structural characteristics in both publications. In the second part of the thesis, electrostatic interactions were used to integrate functional responsiveness in materials. In Publication III, a hydrogel consisting of zwitterionic and nonionic units was demonstrated as a taste-recognizing material. The sensing was based on the interactions between the hydrogel's repeating units and the small-molecular tastant molecules, leading to volumetric and electrical responses depending on the (non)ionic nature of the tastant. In Publication IV, a polyampholyte was used as a ligand to prepare fluorescent gold nanoclusters with pH-responsive photoluminescence, where the protonation of the tertiary amine groups led to enhanced photoluminescence in acidic medium. This feature was further used in the bioimaging of lysosomes. Finally, Publication V demonstrated that a low-magnitude electric field can lead to a controlled locomotion of surfactant-stabilized aqueous droplets. The results suggest that the droplet propulsion occurs through non-equilibrium mechanisms at the zwitterion-covered droplet interface. The results of this thesis contribute to the understanding of ionic interactions and how they can be used in the development of functional responsive materials through equilibrium and nonequilibrium mechanisms.
  • Item
    Naturally Occurring Discursive Work as a Reflection of Organizational Identification During Organizational Transformation
    (Aalto University, 2024) Kupiainen, Olli-Jaakko; Vartiainen, Matti, Emeritus Prof., Aalto University, Department of Industrial Engineering and Management, Finland; Hakonen, Anu, Ph.D, Haaga-Helia University of Applied Sciences, Finland; Tuotantotalouden laitos; Department of Industrial Engineering and Management; Perustieteiden korkeakoulu; School of Science; Vuori, Natalia, Assist. Prof., Aalto University, Department of Industrial Engineering and Management, Finland
    Research on organizational changes traditionally focuses on the outcome of change efforts or change processes. Recent theoretical openings of "work" complement these approaches, which aim to elucidate organizational members' purposeful change efforts. Work has discursive, relational, and material dimensions. This doctoral thesis focuses on discursive work, which emphasizes the role of language in aiming to shape or change an organization's social-symbolic objects. An organization's future, identity, and status represent such social-symbolic objects in this doctoral thesis. An empirical arena of this doctoral thesis is naturally occurring change talk that organizational members generate on the enterprise social media and its discussion board during an organizational transformation. These posts are considered organizational narratives. The strength of the narrative approach is that it acknowledges multiple interpretations of change. This doctoral thesis is a case study consisting of three individual essays and a summary of those essays. Each essay explores the same data through different work lenses: temporal work, organizational identity (OI) work, and status work. The essays show how organizational members integrated their microlevel change talk into their organization's macrolevel change attempts. Essay 1 argues that organizational members engage in future-making by offering solutions and making "if-then" plans to enable their organization to meet its goals in the future. Essay 2 suggests that they discursively construct time- and context-sensitive OIs offering alternative interpretations of ongoing transformation. Essay 3 shows that members engage in status-seeking on behalf of their organization, which is supported or hindered by organizational self-efficacy. This doctoral thesis advances the understanding of organizational change literature by arguing that the discursive "work" in which organizational members engage in on a technological platform can help to set the direction for the organization during transformation when organizational members monitor and assess their organization's interest (one motivational aspect of organizational identification). Thus, it is argued that discursive work and organizational identification are closely linked. The research suggests that organizational members' collective, discursive work through social technologies is a cultural phenomenon in which diverse and critical interpretations of ongoing transformation can also be expressed. Furthermore, discursive work requires resources from the members to make inferences about the situation. Social technologies support discursive work by making multiple interpretations and an organization's change potentials visible via organizational narratives, thus generating a discourse of direction.
  • Item
    Light-matter interactions and topological effects in ensembles of plasmonic nanoparticles
    (Aalto University, 2024) Taskinen, Jani M.; Teknillisen fysiikan laitos; Department of Applied Physics; Quantum Dynamics; Perustieteiden korkeakoulu; School of Science; Törmä, Päivi, Prof., Aalto University, Department of Applied Physics, Finland
    The cornerstone of plasmonics is the fundamental coupling between photons and electron oscillations on the surface of a metal, which allows light to be trapped in subwavelength volumes. The electromagnetic fields of plasmonic excitations are highly confined to the metal-dielectric interface, leading to extreme field enhancement and enabling strong light-matter interactions. Enforcing this effect in carefully designed nanostructures allows the creation of high-quality optical modes that provide efficient coupling between photons and molecular emitters. This dissertation studies lasing and condensation as light-matter phenomena in plasmonic nanoparticle lattices and their underlying connections to topology, which is concerned with properties that are invariant under continuous deformations. The optical modes supported by different nanoparticle structures and the phenomena enabled by them are investigated experimentally: samples are fabricated using an electron beam lithography process, and an angle-resolved spectroscopy setup is used to induce the light-matter effects and to characterize their properties. The experimental methods used in the research are discussed in-depth to enhance reproducibility and to provide tools for future implementations. In Publication I, the polarization and phase properties of a strongly coupled plasmon condensate are studied in a square nanoparticle lattice. Polarization-resolved images of the condensate and its far-field emission pattern are used in a phase-retrieval algorithm. The resulting nonuniform condensate phase is shown to host a topologically trivial winding of the polarization vector, which is treated as a pseudo-spin property of the system. In Publication II, the quantum metric and Berry curvature are measured in a plasmonic lattice constituting the first observation of the quantum geometric tensor in a non-Hermitian system. Nonzero components of the tensor are discovered around high-symmetry lines of the Brillouin zone, explained by the non-hermiticity of the system and pseudospin-orbit coupling. The experimental findings are verified qualitatively by a two-band model. In Publication III, a pattern design method based on the lossy nature of metallic nanoparticles is utilized in creating plasmonic quasicrystals that host modes with polarization vortices. The fabricated samples are combined with dye molecules to demonstrate lasing with extremely high topological charges, verified by polarization-resolved measurements and theoretical considerations.
  • Item
    On the Strategic Importance of Building and Using Complex, Algorithmic Systems
    (Aalto University, 2024) Seppälä, Jane; Tuotantotalouden laitos; Department of Industrial Engineering and Management; Strategy and Venturing; Perustieteiden korkeakoulu; School of Science; Vuori, Timo, Prof., Aalto University, Department of Industrial Engineering and Management, Finland
    Organizations build and use increasingly capable algorithmic systems to solve increasingly complex problems. As algorithmic systems become more powerful, they also become more complex and challenging to understand and manage. Thus, we face a situation where technologies are progressively used in high-stakes situations yet less well understood by traditionally trained managers and organizational scholars. Further, the wide use of such systems has strategic implications: the design of algorithmic systems influences strategic options available, strategy is operationalized into and enacted through such systems, and these systems influence emergent strategy. In this thesis, I explore how organizations build and use complex algorithmic systems and highlight the potential strategic implications of such systems. This thesis consists of three qualitative case studies in which building and using complex algorithmic technologies is at the forefront. Essay 1 explores how technological complexity affects post-merger integration dynamics at a large technology company. Focusing on the perceptions and behaviors of one focal unit in a situation of attempted synergy capture through portfolio streamlining, this study demonstrates how technological complexity affects strategic decision-making and illustrates the bi-directional influence between technological and strategic decisions. Essay 2 studies how a large retail organization develops and uses strategically important algorithmic tools. This study describes how strategy is encoded into the tools through the processes of negotiating a role for the tool, making abstract concrete through quantified measures, and materializing the emerging sense into the tool, and further, how tool use can induce masking of uncertainty by hiding technical details. These processes affect emergent strategy and strategizing dynamics. Essay 3 analyzes how a start-up organization develops and uses advanced speech recognition technology to provide transcription services and products. This study describes how the organization maintained and improved system performance as the system's scope was extended and identifies a set of organizational practices that enable this. This thesis makes three contributions. First, I describe the continuous interplay of the development and use of technology, typically described by organizational literature as two separate processes, thus conceptualizing technology development as an ongoing sensemaking process. Second, I contribute to the literature on strategy-as-practice by detailing the strategic role of complex algorithmic systems and recognizing and describing technology development as a strategic activity. I also portray how strategically important algorithmic systems are built and used and how such systems can influence strategizing dynamics. Third, I contribute to the emerging discussion on data practices by describing the essential and often ill-recognized role organizational processes that generate data and data practices play in building efficient algorithmic systems.
  • Item
    Magnetic field-induced particle assembly and jamming
    (Aalto University, 2024) Liu, Xianhu; Peng, Bo, Dr., Aalto University, Department of Applied Physics, Finland; Teknillisen fysiikan laitos; Department of Applied Physics; Molecular Materials; Perustieteiden korkeakoulu; School of Science; Ikkala, Olli, Distinguished Prof., Aalto University, Department of Applied Physics, Finland
    Ferromagnetic materials possess the ability to undergo magnetization in the presence of an external magnetic field and exhibit rapid responsiveness to magnetic field stimuli, rendering them highly suitable as carriers for stimuli-responsive materials. Furthermore, the assembly of ferromagnetic particles under the influence of a magnetic field allows for the design of assembled superstructures with diverse properties, enabling the fulfillment of specific application requirements. In this thesis, ferromagnetic cobalt (Co) and nickel (Ni) particles with various surface roughness were synthesized utilizing the polyol method. These particles were employed as building blocks for magnetic fieldinduced assembly, resulting in the formation of assembled superstructures with distinct properties. The unique characteristics of these superstructures were systematically characterized, followed by an exploration of their potential applications. Publication I utilizes a relatively smooth-surfaced ferromagnetic Co particles as building blocks enabled the assembly of weakly jammed superstructures under the influence of a magnetic field. The shape of these assembled superstructures could be controlled by adjusting the magnetic field strength, imparting tunable sensitivity in response to pressure stimuli. Publication II employs ferromagnetic Ni particles with comparatively rougher surfaces as building blocks, the heightened interparticle friction reduced the critical packing density required for achieving jamming. Consequently, strongly jammed superstructures were formed under the influence of a magnetic field. The pronounced jamming effect induced structural memory in the assembled superstructures, enabling tunable of jamming through magnetic field application kinetics and ultimately yielding broad tunability in response to stimuli. Publication III demonstrates the transfer of jamming-induced structural memory to electrical signals, which could be further transformed into visible light signals. Additionally, pulsed magnetic fields were employed to modulate the system's responsiveness.
  • Item
    Questions About Learners' Code: Extending Automated Assessment Towards Program Comprehension
    (Aalto University, 2024) Lehtinen, Teemu; Korhonen, Ari, Senior university lecturer, Aalto University, Department of Computer Science, Finland; Tietotekniikan laitos; Department of Computer Science; Learning + Technology Research Group (LeTech); Perustieteiden korkeakoulu; School of Science; Malmi, Lauri, Prof., Aalto University, Department of Computer Science, Finland
    Novice programmers have a limited understanding of the program code they produce. Their programs are often based on code snippets from examples and internet searches. Recently and rather suddenly, artificial intelligence has changed programming environments that can now suggest and complete entire programs based on the available context. However, the ability to comprehend and discuss programs is essential in becoming a programmer who is responsible for their work and can reliably solve problems as a member of a team. Many introductory programming courses have hundreds of students per teacher. Therefore, automated systems are often used to produce immediate feedback and assessment for programming exercises. Current systems focus on the created program and its requirements. Unfortunately, their feedback helps students in iterating toward acceptable code rather than acquiring a deep understanding of the program. This dissertation addresses that gap. The dissertation defines and introduces questions about learners' code (QLCs). After a student has submitted a program, they are asked automated, personal QLCs about the structure and the logic of their program. The dissertation describes a system to generate QLCs and contributes three open-source implementations supporting Java, JavaScript, and Python. The empirical contributions of the dissertation are based on multiple studies that research both quantitatively and qualitatively how novice programmers answer various types of QLCs. From the students, who create a correct program, as many as 20% may answer incorrectly about concepts that are critical to systematically reason about their program code. More than half of the students fail to mentally trace the execution of their program. This confirms that novices' program comprehension needs improvement and instructors may overestimate their abilities. The more students answer incorrectly to QLCs, the more they tinker with their code and have less success on the course. Current artificial intelligence systems respond to QLCs better than the average novice. However, they also lapse into humanlike errors producing failed reasoning about the code they generated, which could present an important learning opportunity for the critical use of AI in programming.
  • Item
    On the connectivity interdiction problem, the geometry of data structures and Eulerian circuits
    (Aalto University, 2024) Obscura Acosta, Nidia; Tietotekniikan laitos; Department of Computer Science; Combinatorics of Efficient Computations; Perustieteiden korkeakoulu; School of Science; Chalermsook, Parinya, Prof., Aalto University, Department of Computer Science, Finland
    Over the last century and since the introduction practical computers, the study of algorithms for optimization problems has become one of the main areas in theoretical computer science. Computer algorithms are nowadays an ubiquitous tool in optimization applications like food production, voting systems, route scheduling, transportation, protein synthesis and drug delivery. However, the progress of these area has faced big theoretical and practical challenges, like the lack of computing resources, the increasing volume of data in big networks and theoretical and structural barriers like NP-hardness. In order to by-pass some of the theoretical barriers, this thesis explores several graph and optimization problems through approximation algorithms, data structures, extremal combinatorics and geometry, establishing new state-of-the-art theoretical results in the following problems: The Connectivity Interdiction Problem. For a given weighted undirected connected graph G and integer k, this problem asks to find the optimal set of edges of cost at most k such that the min-cut of the graph G is minimzed after the deletion of these edges from G. We establish new graph-theoretic structural results relating this problem to a variant of graph cut problems called the Normalized Min-cut problem, which allows us to design new exact and approximate algorithms for the unit-cost case establishing a trade-off between running time and solution quality. Furthermore, we answer an open question from Zenklusen [Zenklusen 2014] by showing that this problem admits an FPTAS (Fully Polynomial-Time Approximation Scheme). The Geometry of the Minimum-cost Online Binary Search Tree Problem. In this problem, we want to find the optimum cost online self-adjusting binary search tree which searches any sequence of keys in the tree, a problem related to the Dynamic Optimality Conjecture [Sleator & Tarjan 1985]. We study the Greedy algorithm as one of the two main candidates for this conjecture in a geometric setting, using the theory of forbidden submatrices and introducing a novel matrix decomposition technique. This allows us to improve known upper-bounds in special cases like the pre-order traversal, deque, split and k-increasing sequences and operations, and to settle completely the conjecture for post-order traversal sequences. Furthermore, we show the NP-hardness of a newly introduced generalization of this problem and efficient approximation algorithms for its general case and special cases. Unique Eulerian Circuits. We study the graph theoretical characterization of directed connected graphs with a unique Eulerian circuit. We show a new characterization of these graphs in terms of cut nodes and degrees of a graph, allowing a simple and efficient algorithm to determine if a given graph has a unique Eulerian circuit. Most importantly, this allows us to characterize and develop efficient algorithms to compute the so called maximal safe solutions for the Eulerian Circuit Problem, a concept arising in bioinformatics applications like genome assembly.
  • Item
    Functional Liquid-Fluid Interfaces Based on Hydrophobin Proteins: An Experimental Study for Medical Applications
    (Aalto University, 2024) Al-Terke, Hedar H.; Paananen, Arja, Dr., VTT Technical Research Centre of Finland, Finland; Joensuu, Jussi, Dr., VTT Technical Research Centre of Finland, Finland; Teknillisen fysiikan laitos; Department of Applied Physics; Soft Matter and Wetting; Perustieteiden korkeakoulu; School of Science; Ras, Robin, Prof., Aalto University, Department of Applied Physics, Finland
    Interfaces are everywhere around us. Any direct interaction occurs at the interface. This thesis explored the potential of functional interfaces to solve various medical application challenges. The four publications presented in this thesis highlight the different ways in which functional interfaces can be utilized to address these challenges. Publication I focused on the relationship between gravity, viscoelasticity, and the shape of water droplets coated with HFBI hydrophobin proteins. By studying the self-organization of hydrophobins at the air-water interface, it was found that a rigid layer is formed at a critical concentration, which affects the droplet morphology. This finding has significant implications for engineering and biomedical applications, as it provides a pathway for controlling the shape of droplets in various systems. Publication II presented a novel antibody extraction method using advanced materials and functional interfaces. By dividing an oil-based ferrofluid into daughter droplets under an external magnetic field, the surface area of liquid-liquid interfaces is increased, allowing for the functionalization and application of these interfaces as a substrate for antibody extraction. Publication III focused on the formation and characterization of protein-coated gas bubbles. By employing a micropipette aspiration technique, the mechanical properties of these bubbles were assessed, and a sealing parameter (Q) was determined to evaluate their gas permeability. These well-characterized bubbles have promising potential as ultrasound-enhanced contrast agents in various biomedical fields. They can be utilized for imaging purposes and targeted drug delivery, opening up new possibilities for medical diagnostics and therapies. Publication IV explored the utilization of hydrophobin protein functionalized bubbles to develop an advanced ultrasound molecular imaging probe. By functionalizing bubbles with a moiety part at their interface, they can attach to specific antigens and reveal diseased cells, such as cancer cells. This innovative approach holds great promise for improving the accuracy and sensitivity of molecular imaging techniques, enabling early detection and precise targeting of diseases. Overall, the findings presented in this thesis demonstrate the immense potential of functional interfaces in solving various medical application challenges. They provide valuable insights into the design and development of novel materials and techniques that can improve diagnostics, therapeutics, and imaging in the biomedical field.
  • Item
    Switchable hydrogel networks based on natural polysaccharides
    (Aalto University, 2024) Eklund, Amanda; Zhang, Hang, Dr., Aalto University, Department of Applied Physics, Finland; Teknillisen fysiikan laitos; Department of Applied Physics; Molecular Materials; Perustieteiden korkeakoulu; School of Science; Ikkala, Olli, Prof., Aalto University, Department of Applied Physics, Finland
    Responsive hydrogels are gaining interest in different applications due to their flexible chemistries, biocompatibility, and softness. This has allowed utilisation in fields such as biomedicine, and electronics. By modifying the microstructure of the hydrogel, different material properties can be introduced and optimised. In this thesis, a natural polysaccharide, agarose, is used to modify the hydrogel network of thermoresponsive polymer, N-isopropylacrylamide (NIPAm) to enhance its properties. Using two different network architectures, the optical and adhesive properties of the hydrogels are controlled using temperature change as a stimulus. In Publication I, agarose is utilised as a primary network that is removed after PNIPAm polymerisation to create channels into the hydrogel. These channels enhance water transportation and enable the hydrogel to undergo phase transitions more quickly compared to traditional PNIPAm. Additionally, the material has a bright white appearance, enabling use in applications such as controllable screens and optical switches. Publication II utilises chemically modified agarose as a macro-crosslinker in the PNIPAm network, producing a hydrogel that shows superior whiteness at smaller thickness of the reflecting layer compared to the channeled PNIPAm. Publication III utilises the water transportation properties of the channeled hydrogel to realise controllable underwater adhesion. Additionally, the hydrogel includes biomimetic catechol groups to enhance adhesive properties. The combination of the adhesion and controllable water transportation allows the adhesion to be switched on and off using a change in temperature with a high switching efficiency, both underwater and in dry conditions. This hydrogel system can be used as a controllable gripper for fragile, lightweight, irregular biological systems as demonstrated, showing the potential of the channeling approach in fields utilising controllable underwater adhesion such as biomedicine and soft robotics.
  • Item
    Empathy Dynamics: A Neuroscientific Perspective
    (Aalto University, 2024) Zebarjadi, Niloufar; Levy, Jonathan, Dr., Aalto University, Department of Neuroscience and Biomedical Engineering, Finland; Neurotieteen ja lääketieteellisen tekniikan laitos; Department of Neuroscience and Biomedical Engineering; Empathy Building Neuro-lab; Perustieteiden korkeakoulu; School of Science; Jääskeläinen, Iiro, Prof., Aalto University, Department of Neuroscience and Biomedical Engineering, Finland
    Empathy, a socio-cognitive process of perceiving the feelings of others, is one of the fundamental basis of healthy social interaction. Advancements in neuroimaging techniques in the past three decades have facilitated a deeper exploration of the neural mechanisms associated with empathy, complementing traditional approaches and broadening the understanding of this complex phenomenon. This thesis investigated the intricate neural basis of empathy and its variations across individuals. In Study I, we employed magnetoencephalography (MEG) to explore frequency-decomposed neural activities during pain empathy. We detected four significant patterns corresponding to different components of empathy including an alpha suppression pattern, two beta suppression patterns, and a late alpha-beta enhancement pattern as well as their link to subjective experiences. In Study II, MEG and Functional Magnetic Resonance Imaging (fMRI) were utilized to examine the maturation of empathy, revealing a shift in neural and functional mechanisms of empathy from adolescence to young adulthood. Studies III and IV delved into the association between political ideology and neural responses to emotional suffering and physical pain of others, respectively, highlighting an intriguing, yet complex, relationship between empathy and political ideology. Overall, the findings in the current thesis advance the understanding of neural processes underlying empathy, underscoring the importance of considering diverse factors such as age, political ideology, and subjective experiences. This research can open new vistas for future exploration, encouraging a more comprehensive approach to neuroscientific investigations of empathy.
  • Item
    Systems and Methods for Multiple-View and Depth-Based People Tracking and Human-Computer Interaction
    (Aalto University, 2024) Korkalo, Otto; Takala, Tapio, Prof. Emer., Aalto University, Department of Computer Science, Finland; Tietotekniikan laitos; Department of Computer Science; Perustieteiden korkeakoulu; School of Science; Kannala, Juho, Prof. Aalto University, Department of Computer Science, Finland; Takala, Tapio, Prof. Emer., Aalto University, Department of Computer Science, Finland
    This thesis presents systems and methods for real-time multiple-view and depth-based optical tracking for specific human-computer interaction and smart environment applications. Multiple-view systems are used for mitigating occlusions, enhancing tracking precision and accuracy, and extending the tracking volume to encompass larger scales. Depth cameras, on the other hand, offer the advantage of directly providing three-dimensional information from the scene, which makes them particularly appealing for spatial analysis. For multi-touch interaction, we developed a tracking approach that utilizes multiple side-view cameras to transform any flat surface into a multi-touch screen. Instead of explicitly triangulating the touch points, we employed an extended Kalman filter-based method in which the states of the touch points are updated whenever an observation is received from any of the cameras, ensuring low latency and rapid update rates. To position the cameras as close to the screen as possible, we employed fisheye lenses with modified distortion model, and explored the optimal camera configuration for achieving robust tracking with varying numbers of cameras and touch points. Accurate intrinsic and extrinsic calibration of cameras and camera systems is essential for optimal data fusion and state estimation. Typically, calibration procedures are carried out manually, which is not only time-consuming but can also be impractical. To address this issue in multiple-view depth camera-based people tracking systems, we have developed an auto-calibration method that directly derives the camera network topology and sensor calibration parameters from observations. Additionally, to account for the uncertainties in the observations during state estimation and data fusion, we developed a measurement noise model as part of the auto-calibration procedure. In mixed reality, the aim of camera pose estimation and tracking is to align the real and virtual environments in real-time and in all three dimensions. To achieve this goal, we developed a computer-aided design model-based depth camera tracking approach that utilizes a fast graphics processing unit-based iterative closest point method for pose estimation. This method can be applied to various objects, as long as a depth map from the object can be generated from the desired viewpoint. We conducted investigations into the applicability and performance of the method with different targets and concluded that the proposed approach exhibits reduced drift compared to simultaneous localization and mapping-based method and outperforms monocular edge-based method in terms of accuracy.
  • Item
    Machine Learning Applications Supporting Large Scale Programming Education
    (Aalto University, 2024) Sarsa, Sami; Hellas, Arto, Dr., Aalto University, Department of Computer Science, Finland; Leinonen, Juho, Dr., Aalto University, Department of Computer Science, Finland; Tietotekniikan laitos; Department of Computer Science; Learning + Technology Research Group; Perustieteiden korkeakoulu; School of Science; Malmi, Lauri, Prof., Aalto University, Department of Computer Science, Finland
    Providing effective individualized education at scale has been a widely explored topic in education research, and the advancement of recent machine learning methods have made it possible to develop increasingly effective adaptive and intelligent learning systems. In particular, the emergence of deep learning models, and most recently large language models, has propelled the educational field forward, providing both new challenges and opportunities for educators. This dissertation addresses some of these challenges and opportunities, focusing on machine learning methods as a means to enhance large scale programming education. We first present methodological considerations for identifying learners at risk of dropping out, and empirical evaluation of modern machine learning approaches for evaluating student mastery of skills. Then, we analyse features that relate to students continuing in a series of open online courses for introductory programming. Relating to the constant need to produce new learning materials to keep course content relevant in the rapidly evolving landscape of programming and computer science, and the fact that producing such mterials with appropriate quality can be a highly time-consuming task for educators, we propose and evaluate a novel approach that leverages large language models to create learning materials, particularly programming exercises and code explanations, which can be personalized for student needs and interests for increased engagement. The approach shows promising results in generating diverse, coherent, and relevant content. Most of the generated exercises were considered sensible, novel, and adhering to given themes and concepts. Further, we evaluate automatically generated code explanations in real educational settings and show that students tend to rate automatically generated explanations useful for their learning, even higher than those of their peers. As means to help students, this dissertation looks into improving the timeliness of feedback, a key aspect in the effectiveness of feedback. This is done through proposing a framework that in-cludes a machine learning step for speeding up automated assessment, which consequently speeds up assessment feedback, and constructing annotated datasets of when and how experts provide feedback and hints to learning programmers that can be used as a reference on when and how future machine learning models or other automated methods should provide feedback. As a whole, the scope of the dissertation encompasses much of the entire educational process, spanning from (1) identifying learners needs and those who would benefit from additional assis-tance, to (2) educators designing content for learning and practice to (3) helping learners through timely and meaningful feedback for learners. The results in this dissertation showcase both methodological issues as well as new avenues for enhancing large scale computing education through machine learning methods.
  • Item
    Computational analyses of transcriptome and DNA methylation data
    (Aalto University, 2024) Malonzo, Maia; Lähdesmäki, Harri, Prof., Aalto University, Department of Computer Science. Finland; Tietotekniikan laitos; Department of Computer Science; Computational Systems Biology Group; Perustieteiden korkeakoulu; School of Science; Lähdesmäki, Harri, Prof., Aalto University, Department of Computer Science. Finland
    Transcriptomics and epigenomics, via DNA methylation, both study regulation of gene expression. Transcriptomics cover both expression of coding genes which lead to protein expression as well as non-coding genes, like microRNAs, which regulate gene expression, as well as alternative gene splicing. DNA methylation, on the other hand, is the addition of a methyl group to DNA, mostly cytosine, that can either repress or enhance gene expression. Analysis of the transcriptome and DNA methylome are performed to better understand development of diseases (and identify biomarkers) and biological processes such as stem cell development. In this thesis, we analyzed transcriptome and DNA methylation datasets to better understand aspects of stem cell regulation and diseases (asthma and Alzheimer's disease) as well as developed methods for analyzing DNA methylation data.Transcriptome analysis was performed on stem cells to elucidate the function of a stem cell-specific gene, POLR3G, and to determine the relationship of the microRNA Let-7 and protein LIN28 in human embryonic stem cells. It was shown that POLR3G functions in stem cell maintenance rather than repression of transcription as most of the differentially expressed genes were downregulated, which included both coding and non-coding genes. Let-7 and LIN28 function in a negative feedback loop in mouse embryonic stem cells. Unlike the assumption that Let-7 and LIN28 function in hESC as in mESC, it was found that both are expressed in pluripotent hESC. DNA methylation analysis was performed in two diseases, Alzheimer's disease and asthma, as well as on stem cells. Blood samples from twins discordant for AD were analyzed. A gene associated with cognitive function, ADARB2, was found to be differentially methylated in both blood and brain samples. Analysis of blood samples from children with atopic and non-atopic asthma and controls showed that genes previously associated with the immune response and asthma, SMAD3 and PTGDS, were found to be differentially methylated in children with atopic and non-atopic asthma, respectively. Analysis of karyotypically abnormal stem cells showed that a gene known to protect cells against DNA damage and oxidative stress, CAT, was found to be hypermethylated. Moreover, CAT was also found to be differentially methylated in publicly available cancer cell line data. Cancer shares the property of self-renewal with stem cells.Two methods were developed for analysis of DNA methylation data. LuxRep identifies differentially methylated loci by modelling the biochemistry of bisulfite sequencing (BS-seq) at the level of individual DNA methylation libraries. It was shown that inclusion of libraries with varying bisulfite conversion rates in methylation analysis increases accuracy of differential methylation detection. LuxHMM uses HMM and Bayesian regression to identify differentially methylated regions and was shown to perform competitively against other published methods.
  • Item
    Machine learning in organizations: The processes of diffusion, capability development, and reframing
    (Aalto University, 2024) Mucha, Tomasz; Seppälä, Timo, University Lecturer, Aalto University, Department of Industrial Engineering and Management, Finland; Tuotantotalouden laitos; Department of Industrial Engineering and Management; Perustieteiden korkeakoulu; School of Science; Gustafsson, Robin, Asst. Prof., Aalto University, Department of Industrial Engineering and Management, Finland
    The commercial diffusion of machine learning (ML) enables the development of novel and previously unattainable organizational capabilities. Over the past ten years, the rapid advances in ML algorithms, hardware, and tooling, increasing availability of data and computing resources, as well as highly publicized implementations of ML by tech giants and other firms have triggered and propelled further the diffusion of ML use in organizations globally. However, despite the relative ease of piloting ML projects, scaling and turning them into ML-based capabilities have proven to be challenging to the majority of organizations. The underlying reasoning behind this difficulty is the fundamentally different nature of ML development and updating. Unlike traditional information technology (IT), ML systems do not require explicit codification of task execution rules and their encoding by the developers into the inferential logic of the system. Instead, ML systems learn from data. This means that the existing approaches to traditional IT development are not enough for organizations to successfully build and keep up to date their ML-based capabilities. Motivated by these challenges and the practical relevance of the problem, the overarching objective of this thesis is to explore how organizations can successfully develop and use ML-based capabilities by uncovering the underlying processes and how they unfold over time. The initial development of such capabilities starts with the organizational adoption of the novel technology. Therefore, to scope out the status of ML technology commercial diffusion, Essay 1 of this thesis explores the extent of ML use by large firms and how it has changed over time. Essay 2 concentrates on the process of ML-based capability development in individual organizations. It uncovers the mechanisms inhibiting and promoting the successful development of organizational capabilities based on ML. Finally, Essay 3 concentrates on the organizational processes required for an ML system to function in novel operating environments or application domains. The third essay, thus, unpacks the process of reframing an existing operational ML system. This thesis contributes both theoretically novel and practically relevant insights into ML diffusion, development, and use by organizations. The promise of the transformative impact of ML – technologies which can learn from data and do not require explicit encoding of rules by humans – can be realized only if we advance our understanding of how organizations can productively harness these technologies. To this end, the thesis extends existing research by assuming an engaged scholarship approach and conducting in-depth longitudinal studies. The insights offered expand the understanding of organizational processes needed for the cultivation of ML-based capabilities beyond their initial development and implementation.
  • Item
    Deep Visual Understanding and Beyond - Saliency, Uncertainty, and Bridges to Natural Language
    (Aalto University, 2024) Wang, Tzu-Jui Julius; Laaksonen, Jorma, Senior University Lecturer, D.Sc. (Tech.), Aalto University, Department of Computer Science, Finland; Tietotekniikan laitos; Department of Computer Science; Content-Based Image and Information Retrieval Group; Perustieteiden korkeakoulu; School of Science; Kaski, Samuel, Prof., Aalto University, Department of Computer Science, Finland
    Visual understanding concerns to what extent a cognitive system can reason about the visual surroundings before it reacts accordingly. While visual understanding is considered crucial, what go beyond are the capabilities of multi-modal reasoning which involve also other modalities. That is, a cognitive system is often faced with a daunting process – how to capitalize on the inputs, usually from one or more modalities – to adapt itself to the world of multiple modalities. More importantly, different machine learning paradigms may be exploited to learn both uni-modal and multi-modal reasoning tasks. This defines the main research question initiating the research endeavour presented in this thesis. In response to the dissertation's core research question, the work provides a number of methods empowered by different machine learning paradigms for both uni-modal and multi-modal contexts. More concretely, it is shown that one can estimate visual saliency, which is one of the most crucial fundamentals of visual understanding, with visual cues learned in an unsupervised fashion. Semi-supervised learning principle is found to be effective in combating class-imbalance issues in scene graph generation, which aims at discovering relationships among visual objects in an image. Moreover, to overcome the primary drawback in vision-language (VL) pre-training and other VL applications, which conventionally necessitate annotated image-text pairs, a novel weakly supervised approach is introduced. Besides, several enhancements have been made to supervised learning applications: Firstly, an improved dense image captioning model is proposed to better exploit different types of relationships between visual objects in an image. Secondly, an enhanced video captioning model is proposed to alleviate the impact brought by the modality gap, which can be commonly found in the widely adopted Transformer models. Lastly, an uncertainty-aware classification model is proposed to learn more robustly under noisy supervision when accounting for data and model uncertainties. These results suggest the usefulness and wide applicability of different learning paradigms. In terms of models' robustness, several breakthroughs have been made and elaborated for both uni-modal and multi-modal applications. The research outcomes encompass numerous findings related to computer vision techniques and their bridges to natural language. The thesis concludes with a discussion on the limitations of each published work and potential future endeavours in both uni-modal and multi-modal research.
  • Item
    Data-driven modelling of human behaviour with complex networks
    (Aalto University, 2024) Takko, Tuomas; Kaski, Kimmo, Senior Advisor, Aalto University, Department of Computer Science, Finland; Tietotekniikan laitos; Department of Computer Science; Complex Systems; Perustieteiden korkeakoulu; School of Science; Lampinen, Jouko, Prof., Aalto University, Department of Computer Science, Finland
    Evolving environments and a growing number of sources for data offer new and interesting possibilities for studying the behaviour of individuals, groups and populations. This data from mobile phones, websites and social media provides opportunities for creating data-driven models where the occurring events, such as pandemics, can be considered as natural experiments in the given system altering the human behaviour therein. In addition to observational data, conducting controlled game experiments with agents and humans can be used for studying micro-level actions and decisions in order to understand the behavioural aspects relevant to emerging sociotechnical systems. Data-driven modelling typically focuses on prediction and explanation of the studied phenomena. Where models with high complexity have been shown to excel in prediction accuracy, interpretable and explainable models are appropriate for studying the complex human behaviour. This doctoral thesis presents data-driven modelling paradigms in studying human behaviour in cooperative games, mobility and cyber space using complex networks. The four research articles focus on interpreting human behaviour and decision-making in the sets of data through the modelling frameworks. The first two publications study the human decision making in a cooperative game with non-overlapping information and the effects from the presence of autonomous agents by conducting two game experiments. First we present a computational model based on probability matching and show that the human perception of risk during the experiment was near optimal while the rationality of choices was not. In the second publication the model is used for agents in a human-agent hybrid experiment. The group composition of humans and agents was shown to affect the game performance and the adaptation to the strategies of the agents with different game objective. The third publication studies human mobility during the COVID-19 pandemic in Finland using aggregated data from mobile phones. We consider the activity data as a set of bipartite networks and investigate projected exposure networks between postal code areas. The projected networks are modelled using gravity and radiation models with population data over the years 2019--2021 and the changes in the networks and model coefficients are analyzed in relation to the pandemic and the related effects of non-pharmaceutical interventions. The model parameters are shown to remain stable before the pandemic and once the pandemic begins they show a correlation to indices of intervention stringency. The final article of this dissertation presents a novel framework for constructing knowledge graphs from unstructured reports of cyber-attacks to create a systemic model for visual analysis for domain experts and for estimating risk in the network of entities connected by their high-level relationships and attributes. We implement the framework pipeline and evaluate the risk measure using a collected set of news reports.
  • Item
    Real-time and sample-efficient learning of computationally rational user models
    (Aalto University, 2024) Keurulainen, Antti; Kaski, Samuel, Prof., Aalto University, Department of Computer Science, Finland; Tietotekniikan laitos; Department of Computer Science; Perustieteiden korkeakoulu; School of Science; Kaski, Samuel, Prof., Aalto University, Department of Computer Science, Finland
    To effectively collaborate with humans, Artificial Intelligence (AI) systems must understand human behavior and the factors influencing it, including their goals, preferences, and abilities. Interactions with humans are typically costly, and in many real-life situations, AI must adapt to human behavior after only a few interactions. Additionally, when AI interacts with humans to learn about their behavior, the interactions need to be conducted without any noticeable delay for the human, which in turn necessitates adaptation in real-time. This thesis investigates how an AI system can learn about other agents in a sample-efficient and real-time manner, using methods based on reinforcement learning. The first contribution of this thesis is a method for learning representations of goal-driven agents' behaviors with neural networks from incomplete observations, showing that they can be used for improving performance in cooperative decision-making tasks. The second contribution concerns the creation of an automated method for producing task distributions and related ground truth data for training a meta-learner to assess the skill level and adapt quickly to the behavior of a cooperating partner. The third contribution presents a novel method for designing informative experiments for estimating the parameters of simulation-based user models without closed-form likelihood functions, and which models are grounded in cognitive science. This method simultaneously amortizes the estimation of these parameters and the designing of experiments. These contributions cover a wide range of settings where useful representations of behavior for improving cooperation are learned, along with the efficient learning of complex user models. The implications of the methods developed, as well as their strengths and limitations, are discussed.
  • Item
    Humans as Information Sources in Bayesian Optimization
    (Aalto University, 2024) Mikkola, Petrus; Tietotekniikan laitos; Department of Computer Science; Aalto Probabilistic Machine Learning Group; Perustieteiden korkeakoulu; School of Science; Kaski, Samuel, Prof., Aalto University, Department of Computer Science, Finland
    Humans are at the heart of the current computational revolution, not only as end-users, but also as integral contributors to computational systems such as machine learning (ML) solutions. This is because these systems depend on data that mainly originate from human activities, such as textual content, artistic creations, or transcribed audio clips. This data is not the only human-derived information flowing into the process, as human expertise plays an important role at all stages of ML development. This thesis reviews methodologies for expert knowledge elicitation, and delves into a promising approach to harnessing humans as a source of information, which is based on the following two ideas. The first idea is to assume the existence of a latent "intuition function" that describes an expert's knowledge over the problem of interest. The intuition function can only be accessed through queries that allow for human feedback, such as preferential queries. Learning the intuition function presents a tractable machine learning problem that can be approached through Gaussian process learning with a probabilistic user model on how the expert data is generated. The second idea pertains to how queries should be selected for an expert and how the expert's knowledge should be applied to the problem of interest. Multi-fidelity Bayesian optimization (MFBO) is a global optimization approach that incorporates multiple information sources with differing levels of accuracy and cost, accelerating the search for optimal solutions. Treating humans as auxiliary information sources within the MFBO framework effectively tackles issues concerning knowledge integration and sample-efficiency. This thesis addresses three problems that arise when humans serve as information sources in Bayesian optimization: (i) the requirement for natural human interaction, (ii) the inherent unreliability of human input, and (iii) the high cost associated with human labor. The articles included in the thesis present novel algorithms as viable solutions to the problems (i), (ii), and (iii). Specifically, we identify problem (ii) as an issue of negative transfer, and we provide an algorithm that establishes theoretical bounds on the negative transfer gap.