15,747 research outputs found
Metadata Representations for Queryable ML Model Zoos
No full textMachine learning (ML) practitioners and organizations are building model zoos of pre-trained models, containing metadata describing properties of the ML models and datasets that are useful for reporting, auditing, reproducibility, and interpretability purposes. The metatada is currently not standardised; its expressivity is limited; and there is no interoperable way to store and query it. Consequently, model search, reuse, comparison, and composition are hindered. In this paper, we advocate for standardized ML model metadata representation and management, proposing a toolkit supported to help practitioners manage and query that metadata.Web Information SystemsHuman-Centred Artificial Intelligenc
A Manifesto of Nodalism
No full textThis paper proposes the notion of Nodalism as a means describing contemporary culture and of understanding my own creative practice in electronic music composition. It draws on theories and ideas from Kirby, Bauman, Bourriaud, Deleuze, Guatarri, and Gochenour, to demonstrate how networks of ideas or connectionist neural models of cognitive behaviour can be used to contextualize, understand and become a creative tool for the creation of contemporary electronic music
The roles of seed banks and soil moisture in recruitment of semi-arid floodplain plants: the River Murray, Australia.
No full textThe decline of floodplain vegetation along the Lower River Murray, South Australia, has evoked recommendations for ‘environmental flows’ to restore and maintain the health of the ecosystem. To assist managers to maximize benefits from environmental flows, this thesis considers the significance of water for germination and recruitment in key floodplain plant species. Three dominant species are considered, including two trees, river red gum (Eucalyptus camaldulensis) and black box (E. largiflorens), and an understorey shrub, tangled lignum (Muehlenbeckia florulenta).
The soil seed bank was dominated by terrestrial annual native plants. Among 1400 seedlings, a single river red gum was found, and no black box or lignum, suggesting that these species do not contribute to the persistent soil seed bank and rely instead on aerial seed banks (serotiny). Sampling of the soil seed bank was continued to determine when seed fall might coincide with appropriate soil moisture conditions. Responses of the soil seed bank to varied water regimes were compared to determine requirements for seedling survival. The results indicated that species richness, rapidity of response and survival time were all promoted by sustained soil moisture.
Stands of eucalypts in various states of health (from very stressed to very healthy) were monitored to identify seasonal patterns in bud crops, flowering, fresh leaves and volumes of seed released from the aerial seed bank. Distinct seasonal phenological patterns were apparent, and suggested alternating flowering among individual trees (biennial for red gum, bi-annual for black box), producing an annual peak in summer. Peak seed rain occurred in summer (December–March) in healthy trees for both red gum and black box, with light seed rain continuing throughout the year. Seed fall from stressed trees was much reduced. Stressed trees responded after a second watering event, with much more varied and extended annual seed fall patterns.
Lignum showed a spring peak in flowering and seed production. There was a prolific response of flowering and seeding to rainfall, but few seedlings survived. Vigorous vegetative growth occurred in existing plants in response to rainfall and watering but no new cloned plants were found during the study. An investigation of chromosomes as a potential tool to appraise the balance between sexual and asexual reproduction in lignum proved inconclusive, although a previous report of octoploidy in lignum was confirmed.
Seeds from all three species and lignum cuttings were tested for their responses to varied watering regimes, based on combinations of simulated rain and flood conditions. The optimal soil moisture for continued growth and survival in all seeds and cuttings was 10 25%, with moisture values <10% causing wilting and death. The results also suggested that red gum and black box seeds which germinate in water under flooded conditions need to be stranded onto moist soil at the water’s edge within 10 days, for the seedling to continue to grow. It was also concluded that germination on rain-moistened soil is a key supplementary mechanism for recruitment, particularly between irregular flood events.
For greatest benefit, the timing of environmental flows should complement any seasonal rainfall and irregular flooding that may occur. Extension of suitable soil moisture conditions (10-25%) for as long as possible after >5 mm rainfall, or after over-bank flows, would increase chances for survival of seedlings. December is the most likely month for maximal benefit from watering in the Lower Murray Valley, for germination and recruitment, based on regional rainfall and flooding patterns.Thesis (Ph.D.) -- University of Adelaide, School of Earth and Environmental Sciences, 200
Optimizing ML Inference Queries Under Constraints
No full textThe proliferation of pre-trained ML models in public Web-based model zoos facilitates the engineering of ML pipelines to address complex inference queries over datasets and streams of unstructured content. Constructing optimal plan for a query is hard, especially when constraints (e.g. accuracy or execution time) must be taken into consideration, and the complexity of the inference query increases. To address this issue, we propose a method for optimizing ML inference queries that selects the most suitable ML models to use, as well as the order in which those models are executed. We formally define the constraint-based ML inference query optimization problem, formulate it as a Mixed Integer Programming (MIP) problem, and develop an optimizer that maximizes accuracy given constraints. This optimizer is capable of navigating a large search space to identify optimal query plans on various model zoos.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.Web Information SystemsHuman-Centred Artificial Intelligenc
The niche of benthic foraminifera, critical thresholds and proxies
No full textEcological studies of benthic foraminifera are carried out to explain patterns of distribution and the dynamics of communities. They are also used to provide data to establish proxy relationships with selected factors. According to niche theory, the patterns of distribution of benthic foraminifera are controlled by those environmental factors that have reached their critical thresholds. For each species, in variable environments, different factors may be limiting distributions both temporally and spatially. For a species or an assemblage to be useful as a proxy its abundance must show a strong correlation with the chosen factor. Since numerous factors influence each species, it is only in those environments where the majority of factors show little variation but one particular factor shows significant variation that the proxy relationship for that factor can be determined. On theoretical grounds, the reliability of using foraminiferal abundance as a proxy of a selected environmental factor should be restricted to the range close to the upper and lower thresholds. For oxygen, foraminifera are potential proxies for the lower limits but once oxygen levels rise to values of perhaps >1 or 2 ml l-1, there is no longer a relationship between oxygen levels and abundance. By contrast, the flux of organic matter over a large range shows a sufficiently close relationship with foraminiferal assemblages so that transfer functions can be derived for the deep sea. However, the relationship at species level is far less clear cut. Much more accurate estimates of primary productivity and modern organic flux rates are required to improve the determination of past flux rates
Influence of hosts on the ecology of arboviral transmission: Potential mechanisms influencing dengue, Murray Valley encephalitis, and Ross River virus in Australia
No full textEcological interactions are fundamental to the transmission of infectious disease. Arboviruses are particularly elegant examples, where rich arrays of mechanisms influence transmission between vectors and hosts. Research on host contributions to the ecology of arboviral diseases has been undertaken within multiple subdisciplines, but significant gaps in knowledge remain and multidisciplinary approaches are needed. Through our multidisciplinary review of the literature we have identified five broad areas where hosts may influence the ecology of arboviral transmission: host immunity; cross-protective immunity and antibody-dependent enhancement; host abundance; host diversity; and pathogen spillover and dispersal. Herein we discuss the known and theoretical roles of hosts within these topics and then apply this knowledge to three epidemiologically important mosquito-borne arboviruses that occur in Australia: dengue virus (DENV), Murray Valley encephalitis virus (MVEV), and Ross River virus (RRV). We argue that the underlying mechanisms by which hosts influence arboviral activity are numerous and attempts to delineate these mechanisms further are needed. Investigations that focus on hosts of vector-borne diseases are likely to be rewarding, particularly where the ecology of vectors is relatively well understood. From an applied perspective, enhanced knowledge of host influences upon vector-borne disease transmission is likely to enable better management of disease burden. Finally, we suggest a framework that may be useful to identify and determine host contributions to the ecology of arboviruses
Australian water markets: trends and drivers 2007-08 to 2011-12
No full textThis report is the third annual companion to the Australian water markets report series.The Australian water markets reports provide information that summarises activity in Australian water markets. It includes information on water market structures. The trends and drivers report: provides an overview of the geographical location of markets and trading mechanismsanalyses trends in volumes and prices of entitlement and allocation trades, with national, southern Murray–Darling Basin (MDB), northern MDB, and outside-MDB analysesanalyses the key market drivers (why is water being traded).Findings The 2011–12 year was the second consecutive above-average rainfall year, following a series of below-average rainfall years from 2001 to 2010. By comparing water market outcomes during wetter and drier periods, the report provides strong evidence that markets are being used by participants as an important business tool for responding to varying seasonal conditions. Improved water storage levels and water availability in the southern MDB in 2011–12 resulted in an increased volume of entitlement trade, while prices declined. The Australian Government continued to be a major participant in the market in the southern MDB as part of its buyback program. Allocation trade has continued to grow every year since 2007–08. Some of the drivers of this increase have however changed over that period of time. Outside the MDB, recent flooding in Queensland and continued low rainfall in Western Australia have been reflected in the levels of trade. Trading in Tasmania is expanding with new irrigation regions being developed. Outcomes in the water market are assessed mainly with reference to prices and volumes traded. Water markets are complex, with many interlinking drivers, however the main driver of water trading activity in the MDB continued to be the positive seasonal and weather conditions. Other drivers of water trade include regulatory and legislative conditions, agricultural market factors and environmental protection initiatives
Building a generalisable ML pipeline at ING
No full textAdvances in data science have caused an increase in the use of Artificial Intelligence (AI), specifically Machine Learning (ML), throughout various fields. Not only in research but in the industry as well, has ML been receiving increasing amounts of interest. Many companies rely on ML models to increase the efficiency of existing processes or offer new services and products. The industry, however, is facing several additional challenges compared to the academic context. One of those challenges is applying the Development Operations (DevOps) model to an ML application, also referred to as MLOps. This thesis sets out to find the specific challenges that practitioners encounter while operationalising ML models. To do so, we perform a single-case case study on an ML pipeline built by the Trade & Communication Surveillance team at the ING bank. This case study consists of conducting a set of interviews and performing a manual code inspection of the pipeline. The team faces challenges ranging from having insufficient time for operationalising each ML project individually to operating in the highlyregulated fintech context. Their pipeline is able to deploy a single ML model but it does not generalise well to other projects. We present the first version of an application that mitigates these challenges. The application is able to deploy ML models to the development environment at ING and can be operated by data scientists to reduce the effort of operationalising an ML model. Computer Science | Software Technolog
'Project smells' - Experiences in Analysing the Software Quality of ML Projects with mllint
No full textMachine Learning (ML) projects incur novel challenges in their development and productionisation over traditional software applications, though established principles and best practices in ensuring the project's software quality still apply. While using static analysis to catch code smells has been shown to improve software quality attributes, it is only a small piece of the software quality puzzle, especially in the case of ML projects given their additional challenges and lower degree of Software Engineering (SE) experience in the data scientists that develop them. We introduce the novel concept of project smells which consider deficits in project management as a more holistic perspective on software quality in ML projects. An open-source static analysis tool mllint was also implemented to help detect and mitigate these. Our research evaluates this novel concept of project smells in the industrial context of ING, a global bank and large software- and data-intensive organisation. We also investigate the perceived importance of these project smells for proof-of-concept versus production-ready ML projects, as well as the perceived obstructions and benefits to using static analysis tools such as mllint. Our findings indicate a need for context-aware static analysis tools, that fit the needs of the project at its current stage of development, while requiring minimal configuration effort from the user. 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.Software EngineeringSoftware Technolog
Audiomobiles, Sculptures and Conundrums
No full textRoberto Gerhard was a pioneer of electronic music in England creating a number of substantial concert, theatre and radio works from as early as 1954. Gerhard’s electronic music is one of the richest repositories for understanding the development of the composer’s late compositional technique. Apart from the Symphony no.3, ‘Collages’, none of Gerhard’s electronic music is published. This paper will discuss aspects of Gerhard’s electronic music, focusing on Audiomobiles (1958-59) and Sculptures (1963)
- …
