69238 research outputs found
Sort by
Adaptive Defense Strategies Using PPPO in Deep Reinforcement Learning for CyberBattleSim
No full textAs modern networks grow more complex, they become increasingly vulnerable to sophisticated
and evolving cyberattacks. The CyberBattleSim platform provides a simulation
environment designed to model dynamic, complex network interactions between
attackers and defenders. This thesis leverages CyberBattleSim to test and validate a
new deep reinforcement learning-based defense framework, which outperforms traditional
defense strategies in terms of network availability and adaptability. While
previous research has primarily focused on improving attacker performance using algorithms
like Rapid Actor-Critic (RAC) and Double Deep Q-Learning (DDQL), or
on utilizing traditional approaches such as credential lookups and Q-learning within
the CyberBattleSim environment. This thesis proposes a new defense framework for
real-time protection against unknown cyberattacks, utilizing deep reinforcement learning
techniques based on the Proximal Policy Optimization (PPO) and Asynchronous
Advantage Actor-Critic (A3C) algorithms. The PPO algorithm ensures stable policy
updates, while A3C facilitates faster adaptation to evolving threats. The framework
employs the CyberBattleSim platform to simulate complex network environments and
evaluate dynamic attack-defense interactions.
Results show that the proposed Parallel Proximal Policy Optimization (PPPO) algorithm
achieves rapid convergence and high cumulative rewards, outperforming Deep
Q-Learning (DQL) and random algorithms. In terms of network availability, PPPO effectively
mitigates attacks, maintaining high availability even with multiple attackers.
Furthermore, the real-time defense evaluation reveals that PPPO has a rapid reaction
time, with a peak defense reaction time of 0.005 seconds under a five-agent setup.
Comparative experiments using CPU and GPU resources show that PPPO benefits
significantly from parallel agents, with GPU-based implementations substantially outperforming
CPU in runtime efficiency. These results underscore the applicability and
efficiency of the PPPO framework in providing adaptive, real-time defense against
dynamic cyberattacks. This thesis advances the use of deep reinforcement learning in
cybersecurity, offering a robust defense mechanism capable of adapting to unknown
and evolving threats in real-time
Elementary abelian subgroups of classical groups
No full textThe main problems considered in the thesis are the following:
(1) Classify the nontoral elementary abelian p-subgroups of the simple algebraic groups
G of types A, B, C and D over an algebraically closed field K of characteristic
different from the prime p.
(2) Identify in G the normalizer and centralizer of a representative of each class of
these subgroups.
(3) Classify the nontoral elementary abelian p-subgroups and identify the normalizer
and centralizer of a representative of each class of the fixed point subgroup G^F
where F is a Steinberg endomorphism of G.
We present the work of Andersen et al. which provides the classification and local structure of nontoral elementary abelian p-subgroups of the Lie group PGLn(C). We classify
the nontoral elementary abelian 2-subgroups of the simple algebraic groups of types B, C
and D over K, and, when it is not feasible to do so, reduce the classification problem to
the associated finite frame group. We either identify in G of types B, C and D the normalizer and centralizer of a representative of each class of these subgroups, or realize the
normalizer and centralizer modulo their connected components in the associated finite
frame group. We classify the nontoral elementary abelian p-subgroups and identify the
normalizer and centralizer of a representative of each class of the finite groups PGLn(q)
and PGUn(q) where p does not divide q, and determine this information for some classes of nontoral
elementary abelian 2-subgroups of PCSp2m(q) where 2 does not divide q
A Crisis of (Environmental) Imagination: Sport and the Limits to Environmental Action in Aotearoa New Zealand
No full textThe materiality of the climate crisis is increasingly difficult to ignore, creating a “code red” for humanity (United Nations Citation2021) that demands “transformational” change (Guterres Citation2020). Climate action has not reflected this urgency. This paper suggests that sport in Aotearoa New Zealand offers distinctive insight into the stuck-ness of climate action under late capitalism. Climate change poses an existential threat to many sports (Breitbarth et al. Citation2023). As a disaster-prone country where both sport and the natural environment are core identity threads, this threat is especially stark in Aotearoa New Zealand. Moreover, the case has an international significance because of Aotearoa New Zealand’s longstanding status as a global social laboratory (Davis Citation2014) where radical and future-leading policy changes have occurred. If radical climate action is to happen, sport in Aotearoa New Zealand is fertile ground. Conversely, environmental policy in Aotearoa New Zealand sport is defined by a fragmented paralysis where there is evident environmental awareness but limited action. This inaction, should not be attributed to local failures but instead distinctly illustrates a secondary dimension of the climate crisis: a crisis of imagination whereby it is difficult to conceive of environmental action outside of the exigencies of capitalism
Axel Honneth’s Theory of Justice: A Critical Evaluation
No full textThis thesis critically examines Axel Honneth’s theory of justice, particularly its foundation in the
concept of social freedom, institutionalised principles of recognition, and the methodology of
normative reconstruction. Honneth’s theory, rooted in a reinterpretation of G. W. F. Hegel’s
Philosophy of Right, departs from liberal traditions by proposing that justice is achieved through the
idea of social freedom embedded in institutional structures that sustain mutual recognition. The
research investigates the practical applicability of social freedom, the adequacy of institutional
recognition in addressing structural injustices, and the methodological robustness of normative
reconstruction in comparison to John Rawls, Nancy Fraser, and Amartya Sen. The findings suggest
that while Honneth’s approach innovatively integrates freedom and justice within democratic
ethical life, it faces certain potential risks, including tendencies towards institutional determinism
and limited critical adaptability. By engaging with complementary perspectives from theorists, such
as Jürgen Habermas, Walter Benjamin, Amy Allen, and Paul Ricoeur, this thesis proposes
refinements for a more responsive and adaptable vision of justice, suitable for addressing complex
injustices in pluralistic societies
The southern initiative: How indigenous values inspire social innovation and impact
No full textIndigenous values are increasingly recognised in helping organisations contribute to wellbeing within and beyond the workplace. Adopting the theoretical lens of Māori economies of wellbeing, this case study examines how The Southern Initiative (TSI), a unit within Auckland Council, incorporates Māori values to co-create place-based solutions and foster whānau (family) wellbeing. Through kōrero (conversations) with three people, a wānanga (collaborative discussion) with TSI members, and analysis of organisational literature, we identified how TSI’s organising approach synthesises social innovation and bureaucracy. We found that indigeneity-embedded intrapreneurship, distributed leadership, and whānau-centred design support TSI’s innovations. Mana (prestige) emerged as a primary organising principle, sustaining TSI’s approach to achieving systemic change. By bridging Indigenous paradigms and conventional managerial practice, this case study demonstrates how Māori values can transform public sector management, elevate social justice, and encourage community resilience. These findings highlight culturally grounded frameworks for delivering social impact and shaping equitable outcomes
Behavioural responses in Gambusia affinis following exposure to emerging contaminants
No full textEnvironmental pollution is escalating rapidly, with emerging contaminants posing significant threats to aquatic ecosystems. To understand these threats, the field of behavioural ecotoxicology is advancing, exploring the complex interactions between contaminants and animal behaviour. Subtle but profound changes in behaviour in fish serve as sensitive biomarkers of toxicity, providing early warnings of sublethal effects that traditional toxicological assessments may overlook. My thesis aims to investigate these nuanced behavioural effects of three such contaminants—17α-ethynylestradiol (EE2), nanoplastics, and 6PPD-Q—on the invasive freshwater fish, the western mosquitofish (Gambusia affinis). Fish were exposed to environmentally relevant levels of each contaminant over 21 days (EE2), 7 days (nanoplastics), and 24 hours (6PPD-Q) and a suite of ecologically relevant behaviours were assessed, including boldness, exploration, anxiety, activity, sociability, feeding, predation, mate choice, and courtship. My findings reveal nuanced behavioural responses in Gambusia affinis to various contaminants. EE2 influenced male mate choice, with control males spending more time with exposed females. Nanoplastics significantly increased boldness, activity, and predation rates in female fish and were able to distribute across multiple organs, including the brain, raising concerns about potential neurological impacts. While exposure to 6PPD-Q led to increased predation rates in female fish. These results suggest that some behaviours are particularly sensitive to contaminant exposure in G. affinis, while others remain stable. Enhanced boldness, increased feeding rates, and altered reproductive behaviours may enhance the invasive success of G. affinis, especially given their preexisting competitive edge over native species. These findings highlight the importance of behavioural studies for improving ecological risk assessments and guiding environmental protection efforts
The Need for a Racial Reckoning in the Education in Emergencies Community: A Focus on Social and Emotional Learning
No full textSocial emotional learning (SEL) is promoted as essential to achieving the education in emergencies (EiE) sector’s dual aims of protection and learning. Yet the diffusion of SEL from the Global North to crisis-affected contexts has largely escaped critical scrutiny. This article interrogates SEL’s contested histories and racialized assumptions, revealing how it often perpetuates white saviorism and reproduces white supremacy by erasing race from both practice and discourse. Much of the literature remains rooted in an ontology of “what works” rather than “what matters,” sidelining community knowledge and structural analysis. Drawing on critical race theory, we examine how SEL implementation in the Global South reinforces racial hierarchies and exploits the vulnerabilities of crisis-affected children. In response, we join wider calls for decolonization, reparation, and antiracist praxis in comparative and international education, offering pathways for reimagining SEL through Indigenous, community-driven, and emancipatory frameworks
Community Displacement: How Architectural Adaptations Show Resistance to Assimilation
No full textThe Republic of the Philippines and the Republic of the Marshall Islands are two pacific island nations that have been the site of centuries of colonisation, agricultural exploitation, and wartime decimation forcing marginalised populations to repeatedly resettle and concentrate in and around urban centres. We trace two of these resettled communities to investigate the objectives of cultural assimilation in contrasting top-down versus bottom-up resettlement morphologies. We conclude by suggesting that adaptations in the resettlement architecture show resistance to the traditions of assimilation, and they are an important possible resurgence of cultural patterns
From ‘Obrigada' to 'Kia Ora': An African-Born Brazilian Immigrant’s Identity Negotiation Journey in Aotearoa New Zealand
No full textThe Occurrence and Consequences of Managerial Misbehaviour: Evidence from China’s Anti-Corruption Campaign
No full textUtilising China’s anti-corruption campaign, launched in 2012, as the context, this thesis explores the intersection of corporate governance, executive corruption, and financial outcomes in Chinese state-owned enterprises (SOEs). It consists of three studies that address three key research questions, examined through the analysis of ownership structures, equity market reactions, and debt market responses, all focused on corporate executive corruption.
The first study examines the role of decentralised ownership structures in fostering executive corruption. Using a dataset spanning 2004 to 2019, the analysis focuses on the multi-layered ownership arrangements of SOEs and their potential to create opportunities for corruption. It begins with comparative t-tests, which show that firms implicated in corruption tend to have more decentralised ownership structures, as measured by the number of organisational layers. Logistic regression and robustness checks further confirm the relationship between decentralisation and executive corruption. To address endogeneity concerns, the study employs a two-stage least squares (2SLS) regression using market governance as an instrumental variable. The findings reveal that decentralised ownership structures, particularly pyramidal arrangements, significantly increase the likelihood of corruption. This study highlights the governance challenges associated with decentralisation and emphasises the need for stronger oversight mechanisms.
The second study shifts the focus to the stock market implications of executive corruption. It investigates how announcements of corruption investigations affect investor perceptions and firm valuation in Chinese SOEs. Using a sample of 96 corruption-related announcements from 2013 to 2017, the study measures the market response using cumulative abnormal returns (CARs). The results reveal significant negative CARs following corruption announcements, reflecting investor concerns over governance failures. Interestingly, the hierarchical level of the implicated executive does not significantly influence the market reaction, suggesting that corruption investigations are perceived as systemic governance failures rather than isolated incidents. Robustness checks validate the findings, underscoring the detrimental impact of corruption on investor confidence and stock performance.
The third study explores the financial and strategic implications of corruption-related forced turnovers of CEOs and Chairmen. This study investigates how firms adjust their financial strategies in response to governance disruptions caused by corruption investigations. The analysis reveals that forced turnovers lead to a significant increase in firms’ cost of debt, reflecting diminished creditworthiness and strained relationships with financial institutions. Additionally, the study examines capital structure adjustments, revealing that firms reduce their reliance on long-term debt and increase their use of short-term debt to manage financial constraints. These findings highlight the effects of corruption on firms’ financial strategies and underscore the importance of trust and reputation in corporate financing.
Overall, these three studies provide a comprehensive analysis of the interplay between governance structures, corruption, and financial outcomes in Chinese SOEs. This thesis contributes to the literature on corporate governance by highlighting the systemic challenges posed by corruption and decentralisation in hierarchical ownership structures. It offers valuable insights for policymakers and practitioners seeking to strengthen governance mechanisms and mitigate the risks of corruption in state-owned enterprises, both in China and in other economies with similar institutional characteristics