5,243 research outputs found

    Scalable Continual Learning Using Cascading Hypernetworks and Cellular Automata

    No full text
    Addressing the critical challenges of catastrophic forgetting and scalability is paramount for developing continual learning systems capable of robust operation in real-world environments. This dissertation presents a novel framework that achieves state-of-the-art performance by integrating two core innovations: a cascading hypernetwork architecture for adaptive model generation and a specialized mini-cellular automata (mini-CA) system for automatic task identification. The proposed cascading hypernetworks employ a hierarchical structure, where parent hypernetworks can generate the weights of other hypernetworks and eventually task networks, enabling efficient learning across an expanding task set. This is coupled with an auto-generative replay mechanism that reconstructs network samples and replays them to the same hypernetwork, significantly mitigating catastrophic forgetting without reliance on large memory buffers. For precise task identification, we introduce mini-CA, an efficient and scalable cellular automata variant tailored for complex spatio-temporal data. By utilizing smaller, specialized units with enhanced parallelism and input scalability, mini-CA surpasses the limitations of traditional CA. The system is adeptly extended to process colored images and sequences and integrates seamlessly with deep learning architectures for multi-channel input. In the unified framework, mini-CA first identifies the incoming task, which then directs the cascading hypernetwork to dynamically generate a tailored, task-specific network. Rigorous evaluations on reinforcement learning and image classification benchmarks demonstrate superior accuracy, latency, and scalability. This research underscores the transformative potential of combining hierarchical hypernetworks with specialized CA to build truly adaptive and continuously learning systems for complex applications.Electrical and Computer Engineerin

    Relational Neurogenesis for Lifelong Learning Agents

    No full text
    Reinforcement learning systems have shown tremendous potential in being able to model meritorious behavior in virtual agents and robots. The ability to learn through continuous reinforcement and interaction with an environment negates the requirement of painstakingly curated datasets and hand crafted features. However, the ability to learn multiple tasks in a sequential manner, referred to as lifelong or continual learning, remains unresolved. The search for lifelong learning algorithms creates the foundation for this work. While there has been much research conducted in supervised learning domains under lifelong learning, the reinforced lifelong learning domain remains open for much exploration. Furthermore, current implementations either concentrate on preserving information in fixed capacity networks, or propose incrementally growing networks which randomly search through an unconstrained solution space. In order to develop a comprehensive lifelong learning algorithm, it seems essential to amalgamate these approaches into a condensed algorithm which can perform both neuroevolution and constrict network growth automatically. This thesis proposes a novel algorithm for continual learning using neurogenesis in reinforcement learning agents. It builds upon existing neuroevolutionary techniques, and incorporates several new mechanisms for limiting the memory resources while expanding neural network learning capacity. The algorithm is tested on a custom set of sequential virtual environments which emulate several meaningful scenarios for intellectually down-scaled species and autonomous robots. Additionally, a library for connecting an unconstrained range of machine learning tools, in a variety of programming languages to the Unity3D simulation engine for the development of future learning algorithms and environments, is also proposed

    Liryczność brulionowa Zygmunta Krasińskiego w kontekście wiersza "Bóg mi odmówił tej anielskiej miary"

    No full text
    Zygmunt Krasiński’s Notebook Lyricism in the context of the poem entitled Bóg mi odmówił tej anielskiej miary The article is an interpretation of Krasiński’s poem with respect to ‘notebook romanticism’ category. The author analyses the poem in the light of Krasiński’s literary creation process, which is preserved in the Romantic art of editing

    Historia mediów polskich w świetle wskaźników bibliometrycznych

    No full text
    [History of Polish Media in Context of Bibliometric Indicators]. The article focuses on answering the question if media history is subject to bibliometric trends in ways similar to natural sciences, or does it have its own peculiar characteristics. The material for research consists of Historiographical Citation Index of Polish Media 1945-2010. Following issues were subject to study, among various aspects characteristic for history: language locality, publishing forms, differences in works' citations depending on its type (contributions, synthesis') and peculiarities of author distributions (Lorenz curve, Lotka law). A half-life indicator is also analyzed, and a HL14 indicator introduced, with related analysis. Comparison and correlation of selected parameters (citations, HL14, Hirsch index, publication number, volume, etc) was produced as well

    Learning to Reframe Problems Through Moral Sensitivity and Critical Thinking in Environmental Ethics for Engineers

    No full text
    As attention to the pervasiveness and severity of environmental challenges grows, technical universities are responding to the need to include environmental topics in engineering curricula and to equip engineering students, without training in ethics, to understand and respond to the complex social and normative demands of these issues. But as compared to other areas of engineering ethics education, environmental ethics has received very little attention. This article aims to address this lack and raises the question: How should we teach environmental ethics to engineering students? We argue that one key aspect such teaching should address is the tendency of engineers towards technical framing of (social) problems. Drawing then on engineering ethics pedagogy we propose that the competencies of moral sensitivity and critical thinking can be developed to help engineering students with problem (re)framing. We conclude with an example from our teaching that operationalizes these competencies.Green Open Access added to TU Delft Institutional Repository ‘You share, we take care!’ – Taverne project https://www.openaccess.nl/en/you-share-we-take-care Otherwise as indicated in the copyright section: the publisher is the copyright holder of this work and the author uses the Dutch legislation to make this work public.Ethics & Philosophy of Technolog

    Body powered thermoelectric systems

    No full text
    Thesis (M. Eng.)--Massachusetts Institute of Technology, Dept. of Electrical Engineering and Computer Science, 2012.This electronic version was submitted by the student author. The certified thesis is available in the Institute Archives and Special Collections.Cataloged from student-submitted PDF version of thesis.Includes bibliographical references (p. 107-111).Great interest exists for and progress has be made in the effective utilization of the human body as a possible power supply in hopes of powering such applications as sensors and continuously monitoring medical devices [1]. This report furthers into the area of thermal energy harvesting, which focuses on using the temperature differential generated between the human body and the ambient environment to generate power. More specifically, a body-powered, thermoelectric-based power supply and system will be introduced and examined, with hopes that this technology will be utilized alongside low-power, medical monitoring applications in order to achieve self-sufficiency. This report also analyzes the performance of existing thermoelectric-based body-powered energy harvesting applications and compares that with the new design introduced in this work. The new designs were able to output upwards of 25[mu]W/cm2 or, equivalently, 280µW for the entire heat sink system. Additionally, this report details the physics associated with thermoelectric modules, addresses the issues with modern thermoelectric heat-sinks, introduces two new types of wearable, conformal heat sinks, quantifies the performance of the body-powered thermoelectric supply, tests a flexible EKG processing board, and analyzes future paths for this project.by Krishna Tej Settaluri.M.Eng

    Influence of Ethylene-1-Alkene Copolymers Microstructure on Thermo-Rheological Behavior of Model Blends for Enhanced Recycling

    No full text
    Polyethylenes (PE) are the most commonly occurring ingredients for post-consumer recyclates (PCR). The structure–property relationships of different types of model PE-based blends are established using multiple thermo-rheological analyses. Although considered “simple”, the complex behavior of model PE-based blends is experimentally demonstrated for the first time for metallocene-catalyzed, linear, low-density polyethylenes (mLLDPE) with different microstructures that are commonly encountered in PCR. During non-isothermal crystallization, the microstructure of mLLDPE predominantly influences the interaction between mLLDPE and LDPE. Based on the mLLDPE microstructure, the molten LDPE phase acts either as a nucleating agent or as a crystallization rate promoting agent. Both rheological and thermal analyses show that higher activation energy is required for the reptation or movement of polymer chains in a highly branched microstructure with long chain branching (LCB) compared to a linear microstructure with short chain branching (SCB). The quasi-melt response, as measured by thermal analysis under non-isothermal conditions, is distinctly different and sensitive to both the SCB and LCB present in the LLDPE/LDPE blends

    Do high and volatile levels of public investment suggest misconduct ? the role of institutional quality

    No full text
    This paper investigates the impact of institutional quality on public investment levels over the period 1984-2008. Moreover, it studies how the volatility of public investment and the quality of infrastructure are affected by institutional quality, and explores the contribution of other critical factors. The findings suggest an inverse relationship between public investment levels and institutional quality, supporting the idea that governments use public investment as a vehicle for rent-seeking or to compensate for the fall in private investment due to the poor business environment. In addition, aid flows, revenues and abundance of natural resources contribute positively to the level of capital spending. The author also finds that high volatility of public investment is associated with a lower quality of governance. An increase in revenues is associated with a reduction in the volatility of capital spending, suggesting that proper macroeconomic management smoothes the investment cycle. Finally, the paper provides some tentative evidence of a positive relationship between institutional quality and the quality of infrastructure.Investment and Investment Climate,Debt Markets,Non Bank Financial Institutions,Public Sector Economics,Emerging Markets
    corecore