Machine Learning for Narrative Generation
Location: Edinburgh, SA
Duration: 5 months
Keywords: Machine Learning, Computer Programming, Problem Solving, Self-motivated, VQA
Defence Science & Technology (DST) is a research organization providing innovative science and technology supporting Defence. It is the second-largest publicly-funded research organization in Australia. This work would be conducted at the DST Edinburgh site on the northern outskirts of Adelaide, South Australia. The Ph.D. must be an Australian citizen, and will be required to obtain a baseline security clearance.
Situational awareness is a key requirement for operators, decision makers, and analysts. In the normal course of their roles this is achieved, in part, by immersion, exploration and manipulation of the information space in order to produce the evidence or products needed to support their actions, decisions, or analysis. This helps establish the context, and determine what is known, what is not known, what is important and what is not important to a particular situation. When automation is introduced to handle the ‘big data’ problems of volume, velocity, and variety, this pathway to situational awareness is largely lost. Storytelling, or narrative, is widely regarded as an effective mechanism for experientially engaging an audience and making sense of complexity, and thus can establish the context needed to achieve situational awareness in these circumstances.
Automated generation of narrative is therefore a possible means of providing users with the information needed to achieve and maintain situational awareness when automation is used to support an operator, decision maker, or analyst. DST is currently conducting R&D looking at various techniques for automated narrative generation to describe complex military situations.
This project will explore the feasibility of using Machine Learning techniques to generate narratives of dynamic situations involving multiple entities interacting with each other, environmental features, and responding to events.
DST is seeking a Ph.D. with knowledge and experience in Deep Neural Networks and associated techniques to identify a suitable formalism for representing the problem, identify what machine learning techniques are appropriate, assess the training requirements, and develop a prototype system that demonstrates the feasibility or otherwise of this approach. This work will be conducted with only broad guidance provided on the research goals, so the Ph.D. is expected to be self-directed and able to progress a body of work under their own initiative.
Research to be Conducted
The objectives for this project are:
- Determine a suitable formalism for representing:
- dynamic situations involving multiple entities interacting with each other, environmental features, and responding to events.
- Narratives describing these situations
- Identify appropriate machine learning techniques and architectures for generating narratives of these situations.
- Assess what training methodologies are appropriate, and what training data would be required for this approach.
We are looking for an intern with the following knowledge, experience, and/or attributes:
- Knowledge and experience with machine learning techniques, in particular knowledge of Deep Learning tools and techniques.
- Computer programming skills, preferably using Python or a similar language.
- Ability to apply scientific principles to define, structure and analyse complex problems and use creativity and initiative in solving these problems.
- Self-motivation and willingness to accept personal responsibility for defining and solving complex scientific problems under limited supervision.
Highly beneficial but not a requirement:
- Knowledge of machine learning techniques as applied to image and video captioning and VQA problems.
*Please give the Business Developer a call if you are unsure if you meet all the necessary requirements.
The expected outcomes of this project are:
- A report or research paper outlining the feasibility of this approach, and a recommended way forward.
- A prototype demonstrating either the feasibility or otherwise of this approach.
- Any recommendations made will be subject to review and further assessment before any business decisions are made on future R&D, and will not be the sole basis of such decisions.
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, spending at least 80% of their time on-site with the industry partner. The intern will be expected to 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.
To participate in the AMSI Intern program, all applicants must satisfy the following criteria:
- Be a PhD student currently enrolled at an Australian University.
- PhD candidature must be confirmed.
- Applicants must have the written approval of their Principal Supervisor to undertake the internship. This approval must be submitted at the time of application.
- Have Australian Citizenship
- Internships are also subject to any requirements stipulated by the student’s and the academic mentor’s university.
26 November 2017
INT – 0367