Efficient HTTP-based Adaptive Streaming of Linear and Interactive Videos PDF Download

Author: Vengatanathan Krishnamoorthi
Publisher:
ISBN:
Category :
Languages : en
Pages :

View

Book Description
Online video streaming has gained tremendous popularity over recent years and currently constitutes the majority of Internet traffic. As large-scale on-demand streaming continues to gain popularity, several important questions and challenges remain unanswered. This thesis addresses open questions in the areas of efficient content delivery for HTTP-based Adaptive Streaming (HAS) from different perspectives (client, network and content provider) and in the design, implementation, and evaluation of interactive streaming applications over HAS. As streaming usage scales and new streaming services emerge, continuous improvements are required to both the infrastructure and the techniques used to deliver high-quality streams. In the context of Content Delivery Network (CDN) nodes or proxies, this thesis investigates the interaction between HAS clients and proxy caches. In particular, we propose and evaluate classes of content-aware and collaborative policies that take advantage of information that is already available, or share information among elements in the delivery chain, where all involved parties can benefit. Asides from the users’ playback experience, it is also important for content providers to min- imize users’ startup times. We have designed and evaluated different classes of client-side policies that can prefetch data from the videos that the users are most likely to watch next, without negatively affecting the currently watched video. To help network providers to monitor and ensure that their customers enjoy good playback experiences, we have proposed and evaluated techniques that can be used to estimate clients’ current buffer conditions. Since several services today stream over HTTPS, our solution is adapted to predict client buffer conditions by only observing encrypted network-level traffic. Our solution allows the operator to identify clients with low-buffer conditions and implement policies that help avoid playback stalls. The emergence of HAS as the de facto standard for delivering streaming content also opens the door to use it to deliver the next generation of streaming services, such as various forms of interactive services. This class of services is gaining popularity and is expected to be the next big thing in entertainment. For the area of interactive streaming, this thesis proposes, models, designs, and evaluates novel streaming applications such as interactive branched videos and multi-video stream bundles. For these applications, we design and evaluate careful prefetching policies that provides seamless playback (without stalls or switching delay) even when interactive branched video viewers defer their choices to the last possible moment and when users switches between alternative streams within multi-video stream bundles. Using optimization frameworks, we design and implement effective buffer management techniques for seamless playback experiences and evaluate several tradeoffs using our policies.

Author: Arian Maghazeh
Publisher: Linköping University Electronic Press
ISBN: 9176851753
Category :
Languages : en
Pages : 81

View

Book Description
Modern embedded systems deploy several hardware accelerators, in a heterogeneous manner, to deliver high-performance computing. Among such devices, graphics processing units (GPUs) have earned a prominent position by virtue of their immense computing power. However, a system design that relies on sheer throughput of GPUs is often incapable of satisfying the strict power- and time-related constraints faced by the embedded systems. This thesis presents several system-level software techniques to optimize the design of GPU-based embedded systems under various graphics and non-graphics applications. As compared to the conventional application-level optimizations, the system-wide view of our proposed techniques brings about several advantages: First, it allows for fully incorporating the limitations and requirements of the various system parts in the design process. Second, it can unveil optimization opportunities through exposing the information flow between the processing components. Third, the techniques are generally applicable to a wide range of applications with similar characteristics. In addition, multiple system-level techniques can be combined together or with application-level techniques to further improve the performance. We begin by studying some of the unique attributes of GPU-based embedded systems and discussing several factors that distinguish the design of these systems from that of the conventional high-end GPU-based systems. We then proceed to develop two techniques that address an important challenge in the design of GPU-based embedded systems from different perspectives. The challenge arises from the fact that GPUs require a large amount of workload to be present at runtime in order to deliver a high throughput. However, for some embedded applications, collecting large batches of input data requires an unacceptable waiting time, prompting a trade-off between throughput and latency. We also develop an optimization technique for GPU-based applications to address the memory bottleneck issue by utilizing the GPU L2 cache to shorten data access time. Moreover, in the area of graphics applications, and in particular with a focus on mobile games, we propose a power management scheme to reduce the GPU power consumption by dynamically adjusting the display resolution, while considering the user's visual perception at various resolutions. We also discuss the collective impact of the proposed techniques in tackling the design challenges of emerging complex systems. The proposed techniques are assessed by real-life experimentations on GPU-based hardware platforms, which demonstrate the superior performance of our approaches as compared to the state-of-the-art techniques.

Author: Måns Magnusson
Publisher: Linköping University Electronic Press
ISBN: 9176852881
Category :
Languages : en
Pages : 75

View

Book Description
Probabilistic topic models have proven to be an extremely versatile class of mixed-membership models for discovering the thematic structure of text collections. There are many possible applications, covering a broad range of areas of study: technology, natural science, social science and the humanities. In this thesis, a new efficient parallel Markov Chain Monte Carlo inference algorithm is proposed for Bayesian inference in large topic models. The proposed methods scale well with the corpus size and can be used for other probabilistic topic models and other natural language processing applications. The proposed methods are fast, efficient, scalable, and will converge to the true posterior distribution. In addition, in this thesis a supervised topic model for high-dimensional text classification is also proposed, with emphasis on interpretable document prediction using the horseshoe shrinkage prior in supervised topic models. Finally, we develop a model and inference algorithm that can model agenda and framing of political speeches over time with a priori defined topics. We apply the approach to analyze the evolution of immigration discourse in the Swedish parliament by combining theory from political science and communication science with a probabilistic topic model. Probabilistiska ämnesmodeller (topic models) är en mångsidig klass av modeller för att estimera ämnessammansättningar i större corpusar. Applikationer finns i ett flertal vetenskapsområden som teknik, naturvetenskap, samhällsvetenskap och humaniora. I denna avhandling föreslås nya effektiva och parallella Markov Chain Monte Carlo algoritmer för Bayesianska ämnesmodeller. De föreslagna metoderna skalar väl med storleken på corpuset och kan användas för flera olika ämnesmodeller och liknande modeller inom språkteknologi. De föreslagna metoderna är snabba, effektiva, skalbara och konvergerar till den sanna posteriorfördelningen. Dessutom föreslås en ämnesmodell för högdimensionell textklassificering, med tonvikt på tolkningsbar dokumentklassificering genom att använda en kraftigt regulariserande priorifördelningar. Slutligen utvecklas en ämnesmodell för att analyzera "agenda" och "framing" för ett förutbestämt ämne. Med denna metod analyserar vi invandringsdiskursen i Sveriges Riksdag över tid, genom att kombinera teori från statsvetenskap, kommunikationsvetenskap och probabilistiska ämnesmodeller.

Author: Jonas Rybing
Publisher: Linköping University Electronic Press
ISBN: 9176853489
Category :
Languages : en
Pages : 115

View

Book Description
Simulations are frequently used techniques for training, performance assessment, and prediction of future outcomes. In this thesis, the term “human-centered simulation” is used to refer to any simulation in which humans and human cognition are integral to the simulation’s function and purpose (e.g., simulation-based training). A general problem for human-centered simulations is to capture the cognitive processes and activities of the target situation (i.e., the real world task) and recreate them accurately in the simulation. The prevalent view within the simulation research community is that cognition is internal, decontextualized computational processes of individuals. However, contemporary theories of cognition emphasize the importance of the external environment, use of tools, as well as social and cultural factors in cognitive practice. Consequently, there is a need for research on how such contemporary perspectives can be used to describe human-centered simulations, re-interpret theoretical constructs of such simulations, and direct how simulations should be modeled, designed, and evaluated. This thesis adopts distributed cognition as a framework for studying human-centered simulations. Training and assessment of emergency medical management in a Swedish context using the Emergo Train System (ETS) simulator was adopted as a case study. ETS simulations were studied and analyzed using the distributed cognition for teamwork (DiCoT) methodology with the goal of understanding, evaluating, and testing the validity of the ETS simulator. Moreover, to explore distributed cognition as a basis for simulator design, a digital re-design of ETS (DIGEMERGO) was developed based on the DiCoT analysis. The aim of the DIGEMERGO system was to retain core distributed cognitive features of ETS, to increase validity, outcome reliability, and to provide a digital platform for emergency medical studies. DIGEMERGO was evaluated in three separate studies; first, a usefulness, usability, and facevalidation study that involved subject-matter-experts; second, a comparative validation study using an expert-novice group comparison; and finally, a transfer of training study based on self-efficacy and management performance. Overall, the results showed that DIGEMERGO was perceived as a useful, immersive, and promising simulator – with mixed evidence for validity – that demonstrated increased general self-efficacy and management performance following simulation exercises. This thesis demonstrates that distributed cognition, using DiCoT, is a useful framework for understanding, designing and evaluating simulated environments. In addition, the thesis conceptualizes and re-interprets central constructs of human-centered simulation in terms of distributed cognition. In doing so, the thesis shows how distributed cognitive processes relate to validity, fidelity, functionality, and usefulness of human-centered simulations. This thesis thus provides a new understanding of human-centered simulations that is grounded in distributed cognition theory.

Author: Le Minh-Ha
Publisher: Linköping University Electronic Press
ISBN: 918075676X
Category :
Languages : en
Pages : 103

View

Book Description
This thesis addresses the need to balance the use of facial recognition systems with the need to protect personal privacy in machine learning and biometric identification. As advances in deep learning accelerate their evolution, facial recognition systems enhance security capabilities, but also risk invading personal privacy. Our research identifies and addresses critical vulnerabilities inherent in facial recognition systems, and proposes innovative privacy-enhancing technologies that anonymize facial data while maintaining its utility for legitimate applications. Our investigation centers on the development of methodologies and frameworks that achieve k-anonymity in facial datasets; leverage identity disentanglement to facilitate anonymization; exploit the vulnerabilities of facial recognition systems to underscore their limitations; and implement practical defenses against unauthorized recognition systems. We introduce novel contributions such as AnonFACES, StyleID, IdDecoder, StyleAdv, and DiffPrivate, each designed to protect facial privacy through advanced adversarial machine learning techniques and generative models. These solutions not only demonstrate the feasibility of protecting facial privacy in an increasingly surveilled world, but also highlight the ongoing need for robust countermeasures against the ever-evolving capabilities of facial recognition technology. Continuous innovation in privacy-enhancing technologies is required to safeguard individuals from the pervasive reach of digital surveillance and protect their fundamental right to privacy. By providing open-source, publicly available tools, and frameworks, this thesis contributes to the collective effort to ensure that advancements in facial recognition serve the public good without compromising individual rights. Our multi-disciplinary approach bridges the gap between biometric systems, adversarial machine learning, and generative modeling to pave the way for future research in the domain and support AI innovation where technological advancement and privacy are balanced.

Author: Anders Andersson
Publisher: Linköping University Electronic Press
ISBN: 9176850900
Category :
Languages : en
Pages : 60

View

Book Description
Development of new functionality and smart systems for different types of vehicles is accelerating with the advent of new emerging technologies such as connected and autonomous vehicles. To ensure that these new systems and functions work as intended, flexible and credible evaluation tools are necessary. One example of this type of tool is a driving simulator, which can be used for testing new and existing vehicle concepts and driver support systems. When a driver in a driving simulator operates it in the same way as they would in actual traffic, you get a realistic evaluation of what you want to investigate. Two advantages of a driving simulator are (1.) that you can repeat the same situation several times over a short period of time, and (2.) you can study driver reactions during dangerous situations that could result in serious injuries if they occurred in the real world. An important component of a driving simulator is the vehicle model, i.e., the model that describes how the vehicle reacts to its surroundings and driver inputs. To increase the simulator realism or the computational performance, it is possible to divide the vehicle model into subsystems that run on different computers that are connected in a network. A subsystem can also be replaced with hardware using so-called hardware-in-the-loop simulation, and can then be connected to the rest of the vehicle model using a specified interface. The technique of dividing a model into smaller subsystems running on separate nodes that communicate through a network is called distributed simulation. This thesis investigates if and how a distributed simulator design might facilitate the maintenance and new development required for a driving simulator to be able to keep up with the increasing pace of vehicle development. For this purpose, three different distributed simulator solutions have been designed, built, and analyzed with the aim of constructing distributed simulators, including external hardware, where the simulation achieves the same degree of realism as with a traditional driving simulator. One of these simulator solutions has been used to create a parameterized powertrain model that can be configured to represent any of a number of different vehicles. Furthermore, the driver's driving task is combined with the powertrain model to monitor deviations. After the powertrain model was created, subsystems from a simulator solution and the powertrain model have been transferred to a Modelica environment. The goal is to create a framework for requirement testing that guarantees sufficient realism, also for a distributed driving simulation. The results show that the distributed simulators we have developed work well overall with satisfactory performance. It is important to manage the vehicle model and how it is connected to a distributed system. In the distributed driveline simulator setup, the network delays were so small that they could be ignored, i.e., they did not affect the driving experience. However, if one gradually increases the delays, a driver in the distributed simulator will change his/her behavior. The impact of communication latency on a distributed simulator also depends on the simulator application, where different usages of the simulator, i.e., different simulator studies, will have different demands. We believe that many simulator studies could be performed using a distributed setup. One issue is how modifications to the system affect the vehicle model and the desired behavior. This leads to the need for methodology for managing model requirements. In order to detect model deviations in the simulator environment, a monitoring aid has been implemented to help notify test managers when a model behaves strangely or is driven outside of its validated region. Since the availability of distributed laboratory equipment can be limited, the possibility of using Modelica (which is an equation-based and object-oriented programming language) for simulating subsystems is also examined. Implementation of the model in Modelica has also been extended with requirements management, and in this work a framework is proposed for automatically evaluating the model in a tool.

Author: Klervie Toczé
Publisher: Linköping University Electronic Press
ISBN: 9180757480
Category :
Languages : en
Pages : 122

View

Book Description
More and more services are moving to the cloud, attracted by the promise of unlimited resources that are accessible anytime, and are managed by someone else. However, hosting every type of service in large cloud datacenters is not possible or suitable, as some emerging applications have stringent latency or privacy requirements, while also handling huge amounts of data. Therefore, in recent years, a new paradigm has been proposed to address the needs of these applications: the edge computing paradigm. Resources provided at the edge (e.g., for computation and communication) are constrained, hence resource management is of crucial importance. The incoming load to the edge infrastructure varies both in time and space. Managing the edge infrastructure so that the appropriate resources are available at the required time and location is called orchestrating. This is especially challenging in case of sudden load spikes and when the orchestration impact itself has to be limited. This thesis enables edge computing orchestration with increased resource-awareness by contributing with methods, techniques, and concepts for edge resource management. First, it proposes methods to better understand the edge resource demand. Second, it provides solutions on the supply side for orchestrating edge resources with different characteristics in order to serve edge applications with satisfactory quality of service. Finally, the thesis includes a critical perspective on the paradigm, by considering sustainability challenges. To understand the demand patterns, the thesis presents a methodology for categorizing the large variety of use cases that are proposed in the literature as potential applications for edge computing. The thesis also proposes methods for characterizing and modeling applications, as well as for gathering traces from real applications and analyzing them. These different approaches are applied to a prototype from a typical edge application domain: Mixed Reality. The important insight here is that application descriptions or models that are not based on a real application may not be giving an accurate picture of the load. This can drive incorrect decisions about what should be done on the supply side and thus waste resources. Regarding resource supply, the thesis proposes two orchestration frameworks for managing edge resources and successfully dealing with load spikes while avoiding over-provisioning. The first one utilizes mobile edge devices while the second leverages the concept of spare devices. Then, focusing on the request placement part of orchestration, the thesis formalizes it in the case of applications structured as chains of functions (so-called microservices) as an instance of the Traveling Purchaser Problem and solves it using Integer Linear Programming. Two different energy metrics influencing request placement decisions are proposed and evaluated. Finally, the thesis explores further resource awareness. Sustainability challenges that should be highlighted more within edge computing are collected. Among those related to resource use, the strategy of sufficiency is promoted as a way forward. It involves aiming at only using the needed resources (no more, no less) with a goal of reducing resource usage. Different tools to adopt it are proposed and their use demonstrated through a case study.

Author: Leif Jonsson
Publisher: Linköping University Electronic Press
ISBN: 9176853063
Category :
Languages : en
Pages : 149

View

Book Description
This thesis investigates the possibilities of automating parts of the bug handling process in large-scale software development organizations. The bug handling process is a large part of the mostly manual, and very costly, maintenance of software systems. Automating parts of this time consuming and very laborious process could save large amounts of time and effort wasted on dealing with bug reports. In this thesis we focus on two aspects of the bug handling process, bug assignment and fault localization. Bug assignment is the process of assigning a newly registered bug report to a design team or developer. Fault localization is the process of finding where in a software architecture the fault causing the bug report should be solved. The main reason these tasks are not automated is that they are considered hard to automate, requiring human expertise and creativity. This thesis examines the possi- bility of using machine learning techniques for automating at least parts of these processes. We call these automated techniques Automated Bug Assignment (ABA) and Automatic Fault Localization (AFL), respectively. We treat both of these problems as classification problems. In ABA, the classes are the design teams in the development organization. In AFL, the classes consist of the software components in the software architecture. We focus on a high level fault localization that it is suitable to integrate into the initial support flow of large software development organizations. The thesis consists of six papers that investigate different aspects of the AFL and ABA problems. The first two papers are empirical and exploratory in nature, examining the ABA problem using existing machine learning techniques but introducing ensembles into the ABA context. In the first paper we show that, like in many other contexts, ensembles such as the stacked generalizer (or stacking) improves classification accuracy compared to individual classifiers when evaluated using cross fold validation. The second paper thor- oughly explore many aspects such as training set size, age of bug reports and different types of evaluation of the ABA problem in the context of stacking. The second paper also expands upon the first paper in that the number of industry bug reports, roughly 50,000, from two large-scale industry software development contexts. It is still as far as we are aware, the largest study on real industry data on this topic to this date. The third and sixth papers, are theoretical, improving inference in a now classic machine learning tech- nique for topic modeling called Latent Dirichlet Allocation (LDA). We show that, unlike the currently dominating approximate approaches, we can do parallel inference in the LDA model with a mathematically correct algorithm, without sacrificing efficiency or speed. The approaches are evaluated on standard research datasets, measuring various aspects such as sampling efficiency and execution time. Paper four, also theoretical, then builds upon the LDA model and introduces a novel supervised Bayesian classification model that we call DOLDA. The DOLDA model deals with both textual content and, structured numeric, and nominal inputs in the same model. The approach is evaluated on a new data set extracted from IMDb which have the structure of containing both nominal and textual data. The model is evaluated using two approaches. First, by accuracy, using cross fold validation. Second, by comparing the simplicity of the final model with that of other approaches. In paper five we empirically study the performance, in terms of prediction accuracy, of the DOLDA model applied to the AFL problem. The DOLDA model was designed with the AFL problem in mind, since it has the exact structure of a mix of nominal and numeric inputs in combination with unstructured text. We show that our DOLDA model exhibits many nice properties, among others, interpretability, that the research community has iden- tified as missing in current models for AFL.

Author: Robert Johansson
Publisher: Linköping University Electronic Press
ISBN: 9179295061
Category :
Languages : en
Pages : 201

View

Book Description
This thesis presents Machine Psychology as an interdisciplinary paradigm that integrates learning psychology principles with an adaptive computer system for the development of Artificial General Intelligence (AGI). By synthesizing behavioral psychology with a formal intelligence model, the Non-Axiomatic Reasoning System (NARS), this work explores the potential of operant conditioning paradigms to advance AGI research. The thesis begins by introducing the conceptual foundations of Machine Psychology, detailing its alignment with the theoretical constructs of learning psychology and the formalism of NARS. It then progresses through a series of empirical studies designed to systematically investigate the emergence of increasingly complex cognitive behaviors as NARS interacts with its environment. Initially, operant conditioning is established as a foundational principle for developing adaptive behavior with NARS. Subsequent chapters explore increasingly sophisticated cognitive capabilities, all studied with NARS using experimental paradigms from operant learning psychology: Generalized identity matching, Functional equivalence, and Arbitrarily Applicable Relational Responding. Throughout this research, Machine Psychology is demonstrated to be a promising framework for guiding AGI research, allowing both the manipulation of environmental contingencies and the system’s intrinsic logical processes. The thesis contributes to AGI research by showing how using operant psychological paradigms with NARS can enable cognitive abilities similar to human cognition. These findings set the stage for AGI systems that learn and adapt more like humans, potentially advancing the creation of more general and flexible AI. Denna avhandling introducerar Maskinpsykologi som ett tvärvetenskapligt område där principer från inlärningspsykologi integreras med ett adaptivt datorsystem. Genom att kombinera forskning från beteendepsykologi med en formell modell för intelligens (Non-Axiomatic Reasoning System; NARS), undersöker avhandlingen hur operant betingning kan användas för att driva utvecklingen av Artificiell General Intelligens (AGI) framåt. Avhandlingen börjar med att förklara grunderna i Maskinpsykologi och hur dessa relaterar till både inlärningspsykologi och NARS. Därefter presenteras en serie experiment som systematiskt undersöker hur allt mer komplexa kognitiva beteenden kan uppstå när NARS interagerar med sin omgivning. Till att börja med etableras operant betingning som en central metod för att utveckla adaptiva beteenden med NARS. I de följande kapitlen utforskas hur NARS, genom experiment inspirerade av operant inlärningspsykologi, kan utveckla mer avancerade kognitiva förmågor som till exempel generaliserad identitetsmatchning, funktionell ekvivalens och så kallade arbiträrt applicerbara relationsresponser. Denna forskning visar att Maskinpsykologi är ett lovande verktyg för att vägleda AGI-forskning, eftersom det möjliggör att både påverka omgivningsfaktorer och styra systemets interna logiska processer. Avhandlingen bidrar till AGI-forskning genom att visa hur operanta psykologiska metoder, tillämpade på NARS, kan möjliggöra kognitiva förmågor som liknar mänskligt tänkande. Dessa insikter öppnar nya möjligheter för att utveckla AI-system som kan lära sig och anpassa sig på ett mer mänskligt sätt, vilket kan leda till skapandet av mer generell och flexibel AI.

Author: Sofia Thunberg
Publisher: Linköping University Electronic Press
ISBN: 9180755747
Category :
Languages : en
Pages : 175

View

Book Description
This thesis explores, through a mixed-methods approach, what happens when companion robots are deployed in care homes for older adults by looking at different perspectives from key stakeholders. Nine studies are presented with decision makers in municipalities, care staff and older adults, as participants, and the studies have primarily been carried out in the field in care homes and activity centres, where both qualitative (e.g., observations and workshops) and quantitative data (surveys) have been collected. The thesis shows that companion robots seem to be here to stay and that they can contribute to a higher quality of life for some older adults. It further presents some challenges with a certain discrepancy between what decision makers want and what staff might be able to facilitate. For future research and use of companion robots, it is key to evaluate each robot model and potential use case separately and develop clear routines for how they should be used, and most importantly, let all stakeholders be part of the process. The knowledge contribution is the holistic view of how different actors affect each other when emerging robot technology is introduced in a care environment. Den här avhandlingen utforskar vad som händer när sällskapsrobotar införs på omsorgsboenden för äldre genom att titta på perspektiv från olika intressenter. Nio studier presenteras med kommunala beslutsfattare, vårdpersonal och äldre som deltagare. Studierna har i huvudsak genomförts i fält på särskilda boenden och aktivitetscenter där både kvalitativa- (exempelvis observationer och workshops) och kvantitativa data (enkäter) har samlats in. Avhandlingen visar att sällskapsrobotar verkar vara här för att stanna och att de kan bidra till en högre livskvalitet för vissa äldre. Den visar även på en del utmaningar med en viss diskrepans mellan vad beslutsfattare vill införa och vad personalen har möjlighet att utföra i sitt arbete. För framtida forskning och användning av sällskapsrobotar är det viktigt att utvärdera varje robotmodell och varje användningsområde var för sig och ta fram tydliga rutiner för hur de ska användas, och viktigast av allt, låta alla intressenter vara en del av processen. Kunskapsbidraget med avhandlingen är en helhetssyn på hur olika aktörer påverkar varandra när ny robotteknik introduceras i en vårdmiljö