Gaining Insights from Complex Suggestions
Engineering, IT, Mathematics and Statistics
- Due to the sensitivity and security of this project, students must have Australian Citizenship to apply. Any applicants not meeting this requirement will automatically be deemed ineligible for this project.
- The Industry Partner has implemented appropriate preparations to comply with Federal and State Government requirements regarding COVID-19 safety. Due to remote arrangements, this internship is now accepting applications from eligible PhD students nationwide.
ABOUT THE INDUSTRY PARTNER
The Science, Technology, Engineering Leadership, and Research Laboratory (STELaRLab) is Lockheed Martin Australia’s national research and development operations centre, and Lockheed Martin’s first non-US based, multidisciplinary R&D laboratory. It is headquartered in Melbourne, with research teams also located in Adelaide and Brisbane. STELaRLab undertakes R&D into information-related technologies: artificial intelligence, machine learning, analytics, image and signal processing, tracking & sensor fusion, and complex optimisation. Lockheed Martin apply those skills to a diverse range of areas including space situational awareness, space operations, multi-domain operational analysis, all-source information fusion, automated information extraction and interpretation, and hyper sonics.
WHAT’S IN IT FOR YOU?
An internship with the STELaRLab will enable you to:
- Gain relevant professional experience working with a distributed and multidisciplinary corporate R&D team
- Work on a significant problem with a direct, meaningful impact
- Perform research in an emerging field spanning computer science, artificial intelligence, and user-centric design
RESEARCH TO BE CONDUCTED
STELaRLab has ongoing research programs relating to how optimisers for combinatorial problems can be used as aids to help reduce the cognitive burden on the user. These problems are such that the space of feasible solutions that need to be considered grows exponentially as the option space gets more complex. Whilst providing optimal recommendations, these outputs can often be counter-intuitive and the underlying reasoning can seem opaque to the user.
A hypothesis has been formulated within the modern field of explainable AI that a human is ultimately seeking comparison with a foil when asking any ‘why’ question. Recently published studies in the literature have demonstrated how such questions of combinatorial optimisers can be viewed as additional constraints. For example, the question ‘why didn’t you take the train?’ made to a route planner can be compiled into a (somewhat obvious) constraint on the optimiser, and the two solutions can be contrasted in terms of travel time, cost, distance covered, and so on. Whilst this process once given a formal constraint is straightforward, the challenge of how to transition from an open question to such constraints, and furthermore what queries a user could provide, is an open question.
STELaRLab is seeking a PhD intern to help drive the design and implementation of a system where a user can explore the reasonings behind a recommendation generated by a software aid. This work, whilst focussed primarily on the computational aspects of the problem, is deeply entwined with other considerations such as natural language, and human-centric design.
The intern will work, under the supervision of STELaRLab staff, to design a system capable of capturing spoken or textual input, compile the query into a form within a known set of optimisable constraints, and finally represent the differences between generated solutions in an intuitive manner. This will involve the formulation and implementation of an optimisation system, a user interface, and potential user surveys.
SKILLS WISH LIST
- Computer science, mathematics
- Experience using natural language processing tools (such as dependency parsers, slot filling, neural models, and so on)
- Experience using modern, high-level languages such as Python
- Adaptability and enthusiasm
- Human-centred computing: human factors, explainable AI, human-machine interfaces
- Experience with off-the-shelf tools for combinatorial optimisation (such as for the travelling salesman problem, postal routing, and so on)
The aim of this project is to develop a proof-of-concept demonstrator which allows:
- A user to provide a spoken or textual query to the system regarding a given output;
- The query to be parsed into a form suitable for further processing;
- A process to be launched to transform the query into a constraint on the optimisation, producing a ‘foil’ solution;
- Visualisation of the differences between the original output and the foil.
The intern will receive $3,000 per month of the internship, usually in the form of stipend payments.
It is expected that the intern will primarily undertake this research project during regular business hours and maintain contact with their academic mentor throughout the internship either through face-to-face or phone meetings as appropriate.
The intern and their academic mentor will have the opportunity to negotiate the project’s scope, milestones and timeline during the project planning stage.
CONNECT WITH APR.INTERN