Postdoctoral Researcher in Machine Learning
An experienced machine-learning researcher is required for an exciting multidisciplinary study of ICT methods for household energy demand reduction, based in one of the UK’s leading machine-learning groups. The post offers the opportunity to work on online and incremental probabilistic inference methods, applied to an unprecedented large and temporally extensive dataset, in a domain that is on a strong upward trajectory with respect to prioritisation and funding.
Start date: 1st April 2013 or as soon thereafter as possible
Duration: 3 years
Salary: £30,122 - £35,938
Benefits: USS pension scheme (worth approx. 16% of salary)
Closing date: 5 p.m. UK time on 14th December 2012
Interview date: week commencing 7th January 2013
See the rest of this website for details about the project.
Informal enquiries about the position to Dr. Nigel Goddard: nigel.goddard at ed.ac.uk or Dr. Charles Sutton: csutton at inf.ed.ac.uk
To apply, visit this link, or in case of difficulty email Nigel Goddard.
Details
An experienced researcher (Postdoctoral Research Associate) is sought for the machine-learning aspects of the £2M IDEAL project – Intelligent Domestic Energy Advice Loop. Based within the Institute for Adaptive and Neural Computation at the School of Informatics, University of Edinburgh, you will join a multidisciplinary team of researchers (in machine-learning, human-computer interaction, sociology and human geography) and social enterprises (Changeworks and the National Energy Foundation) studying new ways to reduce cost and carbon emissions related to household energy consumption. The team will work with a group of around 600 households to gain real-time data on the energy use of the occupants. This data will be collected using non-invasive sensors in dwellings and modelled using machine learning methods. Inference of occupant behaviour will be the basis for computer-generated multimedia feedback to occupants. By providing householders with different types of feedback about their energy use we will evaluate responses to this kind of monitoring, and the effectiveness of tailored feedback. The main goal of the IDEAL project is to construct an enhanced feedback loop which provides information to householders, not just on their energy consumption, but also on what activities they are using energy for, how much energy is used in each activity and what it costs. This information will be matched with suggestions on how to reduce their energy use. The intent is to compare the results with those achieved by simpler feedback on overall energy use, unrelated to behaviours.
The successful candidate will be a probabilistic machine-learning researcher enthusiastic to work in a multidisciplinary team exploring the use of computational methods for domestic energy demand reduction. The Research Associate will focus on methods for inferring behaviours from noisy time-series data, including data cleansing, anomaly detection, feature-engineering and inference in probabilistic models. In the first year this will include helping define the requirements for the sensor system (so that the data supports the necessary inferences), and working with the response generation team members to define what inferences are required for effective feedback to householders. Writing up results for academic publication will be an important aspect of the job.
You will be a self-motivated individual with the ability to take responsibility for key components of the research plan. You will take day-to-day responsibility for the IDEAL Computational Modelling Work Package. You will lead the identification and refinement of research questions, as well as developing innovative methodological approaches. The role will require you to undertake literature reviews, collaborate with diverse project partners, communicate effectively with team members from other disciplines and other areas of Informatics. There are considerable opportunities to shape the details of the research agenda (in consultation with the Lead Investigators and other Postdoctoral Researchers) and to develop new proposals for research funding.
This post offers the opportunity to work in a world-class machine-learning research environment, applying and extending cutting edge methods in an application area (energy) that is on a strong upward trajectory with respect to prioritisation and funding. The dataset gathered and analysed in the project will be unique worldwide in scope and magnitude, and will provide a rich resource for exploring, evaluating and refining relevant machine learning methods.
For a full job specification, visit this link.
Start date: 1st April 2013 or as soon thereafter as possible
Duration: 3 years
Salary: £30,122 - £35,938
Benefits: USS pension scheme (worth approx. 16% of salary)
Closing date: 5 p.m. UK time on 14th December 2012
Interview date: week commencing 7th January 2013
See the rest of this website for details about the project.
Informal enquiries about the position to Dr. Nigel Goddard: nigel.goddard at ed.ac.uk or Dr. Charles Sutton: csutton at inf.ed.ac.uk
To apply, visit this link, or in case of difficulty email Nigel Goddard.
Details
An experienced researcher (Postdoctoral Research Associate) is sought for the machine-learning aspects of the £2M IDEAL project – Intelligent Domestic Energy Advice Loop. Based within the Institute for Adaptive and Neural Computation at the School of Informatics, University of Edinburgh, you will join a multidisciplinary team of researchers (in machine-learning, human-computer interaction, sociology and human geography) and social enterprises (Changeworks and the National Energy Foundation) studying new ways to reduce cost and carbon emissions related to household energy consumption. The team will work with a group of around 600 households to gain real-time data on the energy use of the occupants. This data will be collected using non-invasive sensors in dwellings and modelled using machine learning methods. Inference of occupant behaviour will be the basis for computer-generated multimedia feedback to occupants. By providing householders with different types of feedback about their energy use we will evaluate responses to this kind of monitoring, and the effectiveness of tailored feedback. The main goal of the IDEAL project is to construct an enhanced feedback loop which provides information to householders, not just on their energy consumption, but also on what activities they are using energy for, how much energy is used in each activity and what it costs. This information will be matched with suggestions on how to reduce their energy use. The intent is to compare the results with those achieved by simpler feedback on overall energy use, unrelated to behaviours.
The successful candidate will be a probabilistic machine-learning researcher enthusiastic to work in a multidisciplinary team exploring the use of computational methods for domestic energy demand reduction. The Research Associate will focus on methods for inferring behaviours from noisy time-series data, including data cleansing, anomaly detection, feature-engineering and inference in probabilistic models. In the first year this will include helping define the requirements for the sensor system (so that the data supports the necessary inferences), and working with the response generation team members to define what inferences are required for effective feedback to householders. Writing up results for academic publication will be an important aspect of the job.
You will be a self-motivated individual with the ability to take responsibility for key components of the research plan. You will take day-to-day responsibility for the IDEAL Computational Modelling Work Package. You will lead the identification and refinement of research questions, as well as developing innovative methodological approaches. The role will require you to undertake literature reviews, collaborate with diverse project partners, communicate effectively with team members from other disciplines and other areas of Informatics. There are considerable opportunities to shape the details of the research agenda (in consultation with the Lead Investigators and other Postdoctoral Researchers) and to develop new proposals for research funding.
This post offers the opportunity to work in a world-class machine-learning research environment, applying and extending cutting edge methods in an application area (energy) that is on a strong upward trajectory with respect to prioritisation and funding. The dataset gathered and analysed in the project will be unique worldwide in scope and magnitude, and will provide a rich resource for exploring, evaluating and refining relevant machine learning methods.
For a full job specification, visit this link.
