Alliance One Tobacco (Malawi)
Academic Research Repository at the Institute of Developing EconomiesNot a member yet
52305 research outputs found
Sort by
Analyzing the Effects of Psychedelic Experience on Social Connectedness with a New Measure: The Aliveness Task
No full textThis study investigated the effects of psychedelics on social connectedness by offering a survey to 455 individuals of various demographics. Given previous research and the importance of this topic in mental health realms, this study aimed to provide a new measure focused on social connectedness referred to as the Aliveness Task. With this measure, we asked participants to rate photorealistic images of people, animals, landscapes, and objects. We analyzed data according to five demographics: (1) age, (2) gender, (3) education level, (4) religious views, and (5) previous psychedelic experience. We hypothesized that this task would yield results that reflect those of more established measures used in previous psychological and psychedelic research. While more statistical measures are needed to validate this new measure, we discovered an association between previous psychedelic experience and stronger feelings of connectedness across most demographics.
Participants were recruited using the participant server hosted by the Faculty of Psychology at the University of Zurich and the newsletter of our research group. A self- report questionnaire was used to capture information about participants. This study provides findings that may indicate psychedelics’ significant ability to increase feelings of social connectedness. This study is the first to use this novel measure, adapted by the University of Zurich
Managing Virtualized Network Functions in the Cloud
No full textIn modern networks, network functions (NFs) are extensively deployed to perform specific traffic processing functionalities, such as firewalls, intrusion detection, and cellular signal processing. More and more network operators are beginning to implement network functions in software rather than specialized hardware appliances. This shift brings about advantages, including improved flexibility, increased feature velocity, and mitigated vendor lock-ins. Virtualizing network functions (NFV) is the emerging trend for almost all networks, including ISPs, clouds, and cellular networks.
However, all of these advantages come with trade-offs. First of all, it is not trivial to support high and stable performance for network functions on commodity servers. Single-CPU performance can no longer catch up with network traffic rate, which pushes network functions to distribute their packet processing pipeline across multiple servers. Such distribution requires significant inter-server (and even inter-CPU core) communications, whose overhead becomes scalability bottlenecks for many network functions, such as vRANs. In terms of packet latencies, both software's performance jitters and in-network queuing contribute to unpredictable latencies, significantly impacting service level objectives (SLOs) for many NFs.
Maintaining good performance for network functions is equally critical, including support for performance diagnosis and resilience to small service disruptions. Diagnosing network function performance problems is not trivial, as these problems can be contributed to by many system-level events (e.g., cache misses), and the impact of such events can propagate across network functions and over time. Enabling resilience for network functions includes live upgrades and failovers, which is not trivial due to the high traffic rate and the black-box nature of network functions.
In this dissertation, we propose three key ideas to address these NFV challenges. The first key idea is to use hardware offloading for better performance and predictable latencies, as many new emerging hardware appliances (e.g., SmartNICs, radios) offer more capabilities to help packet processing. The second key idea is to identify critical minimum states inside network functions to support different network function systems. This helps NFV diagnosis accurately pinpoint the root causes and reduces the time spent on state migration for NFV resilience events. Another key idea is to apply domain-specific knowledge for certain network functions (such as vRANs), allowing developers and operators to design optimized pipelines to address scalability bottlenecks and apply domain-specific solutions for efficient state migrations.
To this end, we design four novel NFV systems for different NFV challenges. We first propose Hydra, a scalable distributed massive MIMO system for vRANs, which uses modern hardware radio capabilities to reduce inter-server communication overheads and uses domain-specific pipeline design to reduce inter-CPU core communication overheads, thereby supporting higher scalability. Hydra is the first system to support 150 antennas and 32 users within three servers. We then propose Octopus, a network function to support predictable latencies using SmartNIC offloading. Octopus repurposes the hardware traffic shaping feature on SmartNICs to achieve accurate packet arrival time on the receiving side. Octopus is the first system to support predictable packet latencies within ~50 ns variations. Next, we propose Microscope, an accurate NFV performance diagnosis system. It identifies the critical in-network queuing information for diagnosis, which allows us to accurately pinpoint why a network function suffers from long tail latency issues or packet drops. Microscope achieves 2.5 times higher accuracy than the state-of-the-art solution. Finally, we propose Atlas, a vRAN resilience solution with minimal service disruption. Specifically, it first applies vRAN domain-specific knowledge to identify the critical minimum states for migration, and then repurposes vRAN-specific protocols to help migrate those states without modifying the source codes. Atlas is the first system to enable resilience for vRANs, and it can mitigate service disruptions within a second
Co-targeting Translation Initiation and the RAS/ERK Pathway for the Treatment of KRAS-mutant Lung Cancer
No full textLung cancer is the leading cause of cancer-related death worldwide with non-small cell lung cancer (NSCLC) being the most frequent subtype. Among the genetic alterations associated with NSCLC, activating mutations in the KRAS oncogene account for approximately 30% of cases. Despite extensive research efforts spanning several decades, the development of effective therapies for KRAS-mutant NSCLCs remains limited. While KRAS G12C inhibitors have recently been developed, responses are temporary and are restricted to a subset of patients. In addition, these agents are not relevant for the ~60% of KRAS-mutant NSCLCs that harbor other KRAS mutations. Therapeutic combinations that co- target components of the two major RAS effector pathways (ERK and mTOR) have been shown exert robust anti-tumor effects in animal models. Nonetheless, their clinical application has been limited by toxicity. Therefore, we sought to identify more refined therapeutic targets within these pathways that might selectively affect cancer cells while minimizing damage to normal tissues.
In this dissertation we present compelling evidence supporting the cap-dependent eucaryotic initiation factor 4F (eIF4F) complex as a promising therapeutic target in NSCLC. Importantly, eIF4F is commonly hyperactivated in human cancers due to the convergence of multiple oncogenic pathways on this complex, including RAS/ERK, PI3K/mTOR, and MYC. Its aberrant activation culminates in the increased translation of mRNAs harboring long and highly structured 5’-UTRs, many of which encode proteins that promote tumor growth, development, and drug resistance. Therefore, we hypothesized that suppressing the enhanced translation of pro-tumorigenic mRNAs through inhibition of the eIF4F complex activity might enhance the effects of RAS/ERK pathway inhibitors and selectively kill KRAS-mutant lung cancer cells. Importantly, we found that targeting the eIF4F’s RNA helicase component eIF4A dramatically enhances the effects of KRAS G12C or MEK inhibitors in NSCLCs and induces potent and durable tumor regression in vivo. Moreover, by screening a broad panel of eIF4F targets, we identified pro-survival BCL-2 family proteins as the key translational targets, among many, that mediate the therapeutic response. Finally, we found that concurrent MYC amplification or overexpression dictates sensitivity to these combinations by creating a dependency on eIF4A/F-mediated translation for the expression of the BCL-2 family proteins. Preliminary evidence further suggests that these eIF4A inhibitor- based combinations may exhibit efficacy also against other RAS pathway-driven tumors.
In a separate series of studies, we also found that suppressing the eIF4F complex activity either by targeting the cap-binding protein eIF4E, or by blocking its phosphorylation at Ser209 through the inhibition of MNK1/2 kinases, dramatically enhances the effects of RAS/ERK pathway inhibitors in KRAS- mutant NSCLC cells. These findings further highlight the importance of the eIF4F complex in KRAS- mutant NSCLCs and demonstrate that multiple approaches can be used to inhibit the translation initiation machinery in this tumor type. Interestingly however, while eIF4A and MNK1/2 inhibitors both potentiate the effects of RAS pathway inhibitors, these agents are effective in non-overlapping sets of NSCLC models, suggesting that different signals are responsible for dictating sensitivity to each of these combinations.
Taken together, these findings reveal the potential of targeting cap-dependent translation using diverse strategies to enhance the efficacy of RAS/ERK pathway inhibitors in KRAS-mutant NSCLCs and potentially other RAS pathway-driven tumors. In addition to inspiring the development of specific clinical trials, these findings should encourage further mechanistic investigations into the role of each component of the eIF4F complex in modulating the cell translatome of lung cancer cells. Finally, we hope that our work will fuel additional studies aimed at targeting cap-dependent translation in human cancers, ultimately increasing our understanding of this process in tumor development and the discovery of successful cancer treatments
Economic Lives of Capitalist 'Post-Poverty': Mobility, Vulnerability and Middle-Class Emergence in Indonesia.
No full textThis dissertation examines the emergence of a ‘new’ middle class in Indonesia – as a material and ideological phenomenon. Beginning with an intellectual history of ‘middle-classness’ as a recent indicator of development and poverty alleviation, I argue that the global middle class as a concept has been central to legitimising globally-integrated, capitalist-led growth, particularly as economic orthodoxy has shifted from the crass market-fundamentalism of early neoliberalism towards the increasingly participatory, inclusive and socially conscious paradigms of the ‘post-Washington Consensus’ (PWC). Since the turn of the century, the world’s major finance and development institutions have advocated for poverty alleviation, women’s empowerment and environmental sustainability, but also – more substantively and perhaps even revolutionarily – for labour rights, and for universal healthcare and social protection. Tracing these changes, I argue that perceived middle-class uplift has been central to this ‘new universalism’ and its claims about the mutually reinforcing nature of democratised growth, while demonstrating that these narratives have in fact facilitated and mystified expanding modes of commodification and the production of new, heightened socioeconomic insecurity. Although income and social mobility have become increasingly apparent, so too have new and intensified forms of precarity. Beyond exposing the tenuousness of ‘middle classness’ as a phenomenon in Indonesia and elsewhere, or arguing the inaccuracy and hypocrisy of institutions such as the World Bank and IMF, this dissertation seeks to grapple with the experiences of transformation that have taken place as a result of rapid growth, as well as the effects of mystified commodification on ordinary people’s sense of wellbeing and political possibility
Two strip mall plazas Edison, New Jersey
No full textThrough a suburban counter-ethnobotany, this project examines how plant
and human migrations land in two strip mall plazas in Edison, New Jersey.
Edison became home to large communities of immigrants from East and South
Asia after the 1965 Immigration Act; these communities had culturally
specific needs which they fulfilled through the appropriation of the strip
mall plaza. Plants also inhabit this peripheral asphalt world, both
within the mall and around. Brought by historically complicated global
mechanics, their presence, like that of the people around, is politicized.
As local plant migrations increase due to changes in climate and the
built environment, this project responds by proposing a choreography of
stripping asphalt from the road and parking lot, facilitating planting.
After de-paving, the material and program inside is brought outside and
the spontaneous growth of the back end brought to the front, and the plaza
becomes a garden in migration
Towards Generative Design for Functionality: Topology, Geometry, and Elasticity of Textiles
No full textAs scientists relentlessly seek advancements in the design of function-driven materials for societal benefits such as efficiency and sustainability, generative design of mechanical metamaterials - materials whose properties are primarily defined by internal structures rather than chemical compositions, has attracted significant interest from multi-disciplinary research communities. In this dissertation, I focus on the study of textiles, an everyday object as well as an emerging material system with an analogy to mechanical metamaterials, to embed human-centered functionalities in applications across engineering domains, from wearable devices and soft robotics to compliant architected materials.
The creation of textiles has a multiscale nature, as 1D yarns are interlooped into 2D geometries with topological invariants, and then finally assembled into 3D fabrics. The immense versatility that makes textiles attractive, through iteration of materials at the 1D level and structures at the 2D level, also poses the challenge of predicting their mechanical behaviors as 3D devices, due to design selections made at different length scales. This has created a demand for mechanical and statistical models to investigate the behaviors of textiles across scales, in order to provide physical insights guiding the design of textiles. Meanwhile, the rise of big data has called for the development of suitable parameterization that can complement systematic inverse design techniques for optimized functionalities.
Overall, I adopt a bottom-up paradigm to probe how topology and geometry affect the elasticity of textiles, with first-principle modeling based on 1D yarns as rods. I first detail the generalized description, modeling, and simulation in Chapter 2. Then in Chapter 3, I establish an interdisciplinary paradigm to study weft-knitted fabrics, which are typical building blocks of textile-like metamaterials for their distinctive extensibility without material damage and spatially programmable anisotropy. Using experiments and micromechanically-based simulations, I uncover the roles of yarn dynamics on the nonlinear elasticity of weft-knitted fabrics under uniaxial tension along principal fabric directions. Following that, in Chapter 4, I utilize the validated computational model to further explore the general elastic energy landscapes of weft-knitted fabrics under varying uniaxial and biaxial loading conditions, in order to provide more physical insights into the origin of their nonlinear behavior from macroscopic perspectives. With the learned physical insights on how topology and geometry affect the elasticity of textiles, I propose a yarn-based parameterization for generative design of topologically programmable textiles in Chapter 5. Finally, I draw inspiration from flexible mechanical metamaterials to discuss how to integrate data-driven techniques into the design of textile-like metamaterials and provide future perspectives in Chapter 6
Bioinspired Photonic Crystals: Self-Assembly, Surface Functionalization, and Sensing in Inverse Opals
No full textFrom moth wings to lotus leaves, the natural world has many examples of combining self-assembled nanostructured surfaces and surface energy control to manipulate the interactions between liquids and surfaces.1,2 These functional nanostructures and their assembly pathways can inspire the creation and functionalization of engineered nanostructured materials with widespread applications. In the initial chapters of this thesis, I explore two case studies, where I examine the formation of natural photonic structures in begonia leaves and investigate the potential for shape manipulation in crystals of guanine, which is a commonly used natural photonic material, before moving to fully engineered photonic crystal systems and their applications. Engineers can utilize a different and often more varied palette of materials than those used in nature to optimize for more complex applications, and we can adopt design principles from biological photonic structures to create bioinspired self-assembled nanostructured surfaces over which we have much more control and can customize as an adaptable platform technology. Inverse opals are a particularly interesting platform as they have a tunable interconnected porous geometry that can produce iridescent structural color and can be functionalized with surface-modifying molecules to customize the interactions between the surface of pores and fluids or particles. Inverse opals have been used as structurally colored optical materials,3 catalytic support materials,4,5 and battery electrodes,6 but one application in which they truly shine is as sensors,7–9 where their structural color, porous geometry, and customizable surface chemistry contribute to their ability to function in differentiating between liquids. Several key design elements of inverse opals – namely their pore geometry, surface chemistry control, and photonic performance – can be adapted to use these sensors in different systems. In this thesis, I have explored the self-assembly and functionalization of photonic crystals in natural and engineered materials with a focus on using inverse opals in sensing applications. I have created self-assembled inverse opals with colloids in several size ranges and explored the self-assembly of colloids with silica sol-gel precursors and titania nanocrystals as well as the post processing of inverse opals using atomic layer deposition in order to control the pore geometry, surface material, and refractive index. I have further used several strategies for customizing the surface functionalization: using gradients of silanes to distinguish between concentrations of bile salts for neonatal jaundice testing, connecting further surface modifying molecules to linker silanes to produce dynamic surface chemistries that react to UV light, and selectively functionalizing the sharp features at pore necks for attachment of particles that could enable occlusion-based sensing. Through this research, I have explored the implications of pore geometry and surface energy on the wetting of ordered porous films, introduced dynamic photoswitchable molecules that can change the wettability of the surface in response to light, and explored the possibility of using inverse opals as label-free viral sensors by occlusion-based wetting modulation
INFERNO: Accelerating Inference with Model Compression via Layer-Wise and Task-Specific Pruning
No full textMachine learning inference latency, or the time required to compute an output of a machine learning model, is increasingly becoming a bottleneck in production systems. Compressing models so that there are fewer, or smaller, computations is the dominant paradigm to increase inference throughput; however, existing methods are limited by expensive pre-training or a lack of task-specific fine-tuning. In this thesis, we present INFERNO, a library designed to accelerate inference on tasks for Convolutional Neural Networks (CNN's) and Large Language Models (LLM's) by introducing two novel model compression algorithms. First, for CNN's, INFERNO introduces greedy layer-wise structured pruning, where, given a task, INFERNO results in a >10x reduction in training cost for comparable accuracy or 30% gain in accuracy for a comparable cost, compared to random sampling, and a larger reduction when compared to state-of-the-art (SOTA) data-center pre-training. Second, for LLM's, INFERNO introduces task-specific pruning on a small sample of unlabeled data, resulting in 1-5% accuracy gains with a 50% pruned LLAMA-2 7B compared to existing SOTA LLM pruning methods. We compare work to existing acceleration libraries and reason there is likely plenty of space for further compression
Exploring the Spiritual Value of Slasher Films: Watching Horror Movies as a Sacred Practice
No full textThis thesis describes the author's experience watching the slasher films Black Christmas (1974), Halloween (1978), Scream (1996) and The Cabin in The Woods (2011) as a sacred practice. The rise of the religious "nones" in the United States has necessitated a renewed look at how this population makes sense of the world. The concept of reading as a sacred practice was adopted from Not Sorry Productions and applied to the medium of film. Charles Long's definition of religion as orientation is used to provide a theoretical framework to posit that the practice of sacred engagement with secular texts is a secular equivalent to religious investigation. Horror films have long been part of popular culture and our collective imagination, and their engagement with the monstrous posits them as unique and important vehicles for social commentary, which make them excellent texts for watching as sacred practice. Over the course of an academic year, the author viewed each film multiple times with different participants for each viewing and recorded the subsequent meaning making conversations. It was found that individual horror films offer virtually endless opportunities for discovery, which facilitates a deeper understanding of the world in which we find ourselves and what it means to be human. These meanings can only be uncovered through repeated viewings combined with intentional conversations with others
The Pattern of Life— Discourses on Life in Literature, Art, and Design of Modern Japan
No full textThis dissertation reconsiders Japanese modernism and modernity through the discouses on life. It examines the discourse on life and how it impacted modern cultural production in Japan from 1911-1941. The idea of life was shaped by evolutionary biology and the philosophies of life flourished around the turn of the 20th century. This dissertation traces the idea of life, and reveals that not only the physicists, but also philosophers, architects, novelists, and designers conceived biological life as material patterns.
The intimate link between life and patterns became prominent in the interwar period, but it started to emerge in the late 19th century, not just in Japan, but in Europe and America as well. For instance, Charles Darwin struggled with the dizzying patterns of the peacock’s tails, which inspired his alternative evolutionary theory of biological life that differs from natural selection, namely, sexual selection. This dissertation also examines the aesthetic and political implications for the idea of life based on the alternative evolutionary theories, such as Natsume Sōseki’s aesthetic evolution, and Itō Chūta’s evolution of architecture. As such, I investigate how the ways in which they were embedded within the global colonial and imperial conditions.
In addition, this dissertation focuses on case studies that link patterns and ideas of life together during the interwar period. For instance, the philosopher Kuki Shūzō considered the striped pattern the expression of the life energy of the Japanese race. Designer and sociologist Kon Wajirō recorded metropolitan data, such as the clothing patterns and decorative patterns of urban housing, as the materialization of mass life. Architect and historian Itō Chūta conducted tran-Eurasian fieldwork research on a pattern, karakusa, based on biological approach. Physicists like Terada Torahiko considered life as crack patterns—the distribution of life energy on uneven fields. The Department Store Takashimaya hired poets and designers to produce and advertise their kimono patterns of life.
Ultimately, across these case studies, I claim that in the interwar period Japan, biological life (seimei) was conceived to be a material, and sometimes decorative, pattern (moyō), that appeared and arranged on the surface of a field. At the same time, patterns were considered to be a physical phenomenon of the formation and growth of life, just as the patterns were also a social-political construct that ordered the space-time of modernity