Search

The principal aim of this dissertation is to examine the dynamics of the contemporary reconstruction of the Ethiopian state on the basis of "ethnic federalism" and democratisation. Among the many African multiethnic polities, Ethiopia is the only country that recognises ethnicity explicitly as an organising principle, even to the extent of de-emphasising the idea of a unitary state and national identity. According to Article 39 (1) of the new Ethiopian Constitution, "every nation, nationalities and peoples in Ethiopia has a right to selfdetermination, including secession. " No other constitution (except, briefly, the former Soviet Constitution) has ever gone so far as to allow such a right. This is perhaps understandable in view of the recent history of regional and ethno-political violence in Ethiopia. The study adopts an historical approach using a qualitative methodology. It analyses and demonstrates how the policies of state centralisation and Amhara hegemonic control transformed ethnic identity into nationalist mobilisation and conflict that finally ended military rule and brought about the demise of Amhara hegemony. It then examines the government that replaced the military regime of Mengistu, its theoretical and philosophical underpinnings of "ethnic federalism" and the democratic accommodation of diversity. The study finds that Ethiopia's political system that began in 1991 has successfully established a new federal democratic order but has so far failed to become a truly "selfgoverning unit" and consolidated democracy. Democratisation and devolution of power opened channels for Ethiopian nations and nationalities to participate as equal citizens in Ethiopian political life and to access to political power, resources, and to protect their ethnic interests both at the national and local levels. It resulted in the formation of local autonomy where regional states were set up on the basis of ethnicity. Ethiopian political change has taken place under inauspicious circumstances that are generally unfavourable to democratic transition and consolidation. And, whilst the EPRDF has made major strides towards successful democratisation in spite of these conditions, it has been unable to consolidate fully the new federal institutions in Ethiopia. As a result, it has instead been transformed into a pseudo-federal and democratic state with minority [Tigrean] hegemony at the centre. The study concludes that non-democratic federalism, with which the EPRDF regime tried to experiment, can generate violence rather than serving as a political panacea for ethnic conflict, as also attested by Yugoslav and Soviet experience. The study stresses that successful federalism requires the end of TPLF hegemony and a democratic arrangement that can facilitate "real self-government" for the nation, nationalities and peoples of Ethiopia in line with the new Ethiopian Constitution. It notes that a democratic mechanism is effective as a means of dealing with ethnic cleavages in plural societies like Ethiopia. Thus, only if Ethiopia's democratisation can truly progress, can its political integration also advance and support for separatist movements consequently weaken If not, ethno-political conflict will continue and at worst the Yugoslav scenario might follow. Clearly the study of Ethiopian efforts at democratic change has relevance for similar problems beyond its boundaries. Recent events around the world have shown that nationalist conflicts are an important feature of the post-Cold War World. Although not a completely new phenomenon, ethnic conflicts are considered crucial challenges to national and international politics alike and are often accompanied by a gradual collapse of state authority, particularly in Africa. The persistence of ethnic identity in developed societies as well as in the former communist and developing states has challenged theories that assume that ethnic identity would disappear through modernisation. The ongoing civil war in many countries illustrates the problems of ethno-political conflicts and the needs for its management. Democratisation and power-sharing is emerging as a key element in contemporary [post] civil war settlements and to manage conflict in deeply divided societies.

1. Rainbow trout (Salmo granderi Richardson) was shown to possess lipoprotein lipase which was released into the circulation after intravenous injection of heparin, reaching maximum activity after 40 minutes. 2. Trout post-heparin plasma LPL showed maximum activity at a temperature of 35°C and pH 8.2. Sodium chloride, protamine sulphate, sodium deoxycholate and heparin showed inhib similar to that of human LPL (EC 3.1.1.3) and distinguishes it from hepatic salt-resistant lipase (SRL) which was not detected in trout post-heparin plasma. 3. Heparin-Sepharose affinity chromatography of trout post-heparin plasma resulted in a peak of LPL activity and no peak was obtained for SRL. The purified LPL showed high specific activity which was increased to several times than that of post-heparin plasma activity. 4.The effects of specific inhibitors on the purified enzyme was the same as obtained with trout post-heparin plasma. 5. The purified enzyme was shown to be serum dependent. A high stimulatory effect was obtained with trout serum VLDL, though stimulation was also obtained with HDL and LDL. The dependancy of the activity on serum gave evidence that the purified enzyme is serum stimulated lipase. iftory effects on the activity; this behaviour is. 6. The amounts of FAs liberated by the enzyme hydrolysis were proportioned to the incubation time. The rate of hydrolysis increased linearly with the amount of the enzyme. Increasing the amount of substrate resulted in a saturating of the enzyme, suggesting that. Michaelis-Mentlen kinetics were operative; an apparent of approximately Km was obtained. 7. Polyacrylamide gel electrophoresis of the purified enzyme showed only a single band. SDS polyacrylamide electrophoresis illustrated that the enzyme was composed of only one polypeptide with an apparent molecular weight of 63,000. 8. Different trout tissues showed low but variable activity; no activity was detected in some tissues but the highest activity was obtained in adipose tissue. 9. The highest activity from acetone-ether powder was by extraction with ammonia/HCl buffer containing sodium deoxycholate and potassium oleate. Heparin and glucose in the same buffer and also NaCl in sodium barbital buffer were less effective as extractants. 10. Adipose tissue activity was inhibited by the specific inhibitors, though little stimulation was obtained by heparin. Fresh tissue activity was stimulated by serum. These observations indicated that the enzyme in adipose tissue is similar to that of LPL in mammals. In the absence of serum acetone-ether powder, activity was increased by NaCl in a manner characteristic of SRL. 11. Heparin-Sepharose chromatography of adipose tissue produced two isolated activities, one was stimulated by NaCl and the other was inhibited by NaCl and stimulated by either serum or VLDL. 12. The presence of LPL in trout, with similar characteristics to those in mammals, strongly support the suggestion that fish transport and take up lipid by mechanisms analogous to those in mammals.

With numerous breakthroughs over the past several years, deep learning (DL) techniques have transformed the world of artificial intelligence (AI). The abilities that were once considered unique and humane, are now characteristics of powerful machines. State-of-the art performance across various perceptual tasks from computer vision, speech recognition, game playing, and others, have been demonstrated. Now that we know it works, current research is more directed towards exploring: (a) TinyAI: how to make it more efficient (deployable in resource-constrained devices), through developing new optimization algorithms and tooling; (b) AutoAI: how to reduce human effort and speedup the development cycle of AI systems through automation. (c) InterpretableAI: understand why it works, through detailed theoretical studies; (d) AppliedAI: how to combine all these efforts to move from “Narrow AI” (resolving specific task) into “General AI” (human semi-equivalent).This dissertation primarily takes on the exploration of the first and second directions (autoAI & tinyAI). In particular, we make progress towards developing algorithms for more efficient and automated AI systems with particular focus on quantization methods.(i) Discovering optimal quantization bitwidths. Research question: What is the optimal bitwidth per layer for optimal quantization of a deep neural network? Proposal: we devel- oped a systematic approach to automate the process of discovering the optimal bitwidth for each layer of a deep neural network while complying to the constraint of maintaining the accuracy through an end-to-end deep Reinforcement Learning framework (ReLeQ). (ii) Quantization-aware training. Research question: Can we train a DNN in such a way that makes them inherently robust to quantization? Proposal: we developed a novel quantization- friendly regularization technique based on sinusoidal function, called WaveQ. WaveQ exploits the periodicity, differentiability, and the local convexity profile in sinusoidal functions to automati- cally propel weights towards values that are inherently closer to quantization levels. Moreover, leveraging the fact that sinusoidal period is a continuous valued parameter, we utilized it as an ideal optimization objective and a proxy to minimize the actual quantization bitwidth, which avoids the issues of gradient-based optimization for discrete valued parameters. (iii) Improved and accelerated finetuning methods. Research question: Can we finetune a quantized DNN in an efficient way to better improve its final accuracy? Proposal: we developed a novel finetuning algorithm for quantized DNNs. The proposed approach utilizes knowledge distillation through teacher-student paradigm in a novel setting that exploits the feature extraction capability of DNNs for higher-accuracy quantization. This divide and conquer strategy makes the training of each student section possible in isolation while all these independently trained sections are later stitched together to form the equivalent fully quantized network.

CO2 emissions from the transport sector are expected to rapidly increase over the coming years, especially in developing countries. The literature cites numerous measures for mitigating transport CO2 emissions. Diverse technocratic, participatory and integrated approaches and methods exist for assessing these measures. However, studies that explore the feasibility of their implementation in developing countries are still insufficient. Further, assessment methods designed specifically for developing countries characterised by top-down policymaking are required. Participation of stakeholder groups and the general public in the assessment of transport CO2 mitigation measures is considered essential. However, incorporation of their perspectives and preferences within the assessment process in general, and within developing countries in particular, is still needed. Further, while the Fifth Assessment Report of the Intergovernmental Panel on Climate Change emphasises the need to understand public views and preferences in developing countries, and how they can be influenced by social learning, empirical evidence in this regard is lacking. This research consequently develops an integrated model for assessing transport measures in relation to climate change mitigation in Bahrain, a high-income, oil-exporting, developing country. It also aims to advance understanding of stakeholders’ views and preferences regarding mitigation measures and provide empirical evidence on the impact of social learning on assessment results. Several assessment methods were embedded within a multi-criteria analysis methodology that applied the analytic hierarchy process (AHP). These included environmental and economic assessments, scenario analysis and participatory techniques. A consultative interview-based model and a deliberative workshop-based model were built for assessing the feasibility of implementing mitigation scenarios in Bahrain and exploring the impact of social learning. The research results indicate that applying fuel economy standards can potentially reduce emissions by 22%. A higher percentage (37%) is attainable using a scenario packaging approach. Results obtained from the first assessment model suggest that no significant differences exist between the preferences of policymakers and experts and those of the general public. Scenario packages were developed as no dominant mitigation scenario was observed. These enabled an exploration of alternatives to lower carbon mobility in Bahrain that were politically and socially acceptable, environmentally effective and economically feasible. Empirical findings on social learning suggest gains in participants’ knowledge and an increased acceptance of a taxation policy. However, the resulting priorities did not differ significantly from those obtained using the first model. The methodological originality of this research lies in its three-way extension of the AHP methodology. Multi-AHP models accommodate divergent feedback from participants. Scenario packages enable benefits from desired scenarios to be merged. Last, examining the plausibility of the assessment results ensures acceptability and successful implementation. While the research findings may be case-specific, lessons learnt can be transferred to other contexts. First, the analysis proves that emission reduction potential exists and can be achieved through the implementation of various policies. Second, abatement costs can be relatively low despite fuel price distortions. Third, incorporating stakeholders’ preferences during the assessment process can increase acceptance of mitigation scenarios. Social learning can also contribute to this. Fourth, applying an integrated model using AHP for assessing transport CO2 mitigation measures in developing countries can be rewarding. This model’s application within other energy-consuming sectors, and possibly in developed countries, may also prove fruitful.