Monash University Indonesia Inspire Doctoral Scholarship Program (IIDSP)

Welcome to the Monash Indonesia Inspire Doctoral Scholarship Program! We are pleased to announce the opening of the July 2023 scholarship round. This program offers an incredible opportunity for talented students to pursue their PhD studies with Monash University, Indonesia. The scholarships are highly competitive and are offered to applicants with an excellent research and/or academic track record.

PhD Scholarship Overview

• Scholarship Duration: 3.5 years
• Scholarship Rounds: July and January each year
• Scholarship Value: The scholarships cover PhD tuition fees, project-related costs, and provide an annual living stipend of AUD $10,632.

Timeline

• Applications Open: 10 July 2023
• Application Deadline: 28 August 2023
• Successful Applicants Contacted: 11 September 2023

Table of Contents

• Data Science. Dr Taufiq Asyhari, Explainable AI for Mission-Critical Applications via Collaborative Human-Machine Time-Sensitive Generated Feedback
• Data Science. Dr Derry Wijaya, Extending Large Language Models (LLMs) capability for regional and low-resource languages
• Data Science. Dr Arif Perdana, Building Bridges: Safeguarding Ethical and Legal Boundaries in the Human-AI Coexistence Era
• Data Science. Dr Risqi Saputra. Spatio-Temporal Deep Learning for Semantic Segmentation of Mining Footprints using Multispectral Satellite Imagery
• Business Innovation. Dr Yessy Arnold Perangin Angin, The Implementation of Environmental, Social and Governance (ESG) Principles and its Impact
• Cybersecurity. Dr Rizka Purwanto. Secure AI-Enhanced Light Curve Analysis for Satellite Detection in Space Surveillance
• Cybersecurity. Dr Muhamad Erza Aminanto. Enabling Privacy-Preserving Artificial Intelligence in Smart City Context Surveillance
• Public Policy and Management. Dr Altaf Virani. Governance and Policy Design Challenges of Public Sector Reform in Indonesia

Supervisors and Project Descriptions

Data Science

Professor Taufiq Asyhari

Project Title: Explainable AI for Mission-Critical Applications via Collaborative Human-Machine Time-Sensitive Generated Feedback

Project Description: Past decades have witnessed massive growth of artificial intelligence (AI) enabled systems across multiple sectors: From IT and engineering to healthcare and environmental monitoring; From financial market and ecommerce to social enterprises. Despite its recent successes, a vast majority of AI-based systems, particularly those relying on Neural Network construction (e.g., various deep-learning architectures), is considered a “black-box” in the sense that although accurate performance in the model development is considerably high, the underpinning mechanics of algorithmic decision remains largely less understood by human. This, to some extent, hinders conscious public acceptance of its wider deployment and scaling up due to the fears of unknown “black-box” making critical decisions for them.

This project aims to investigate the development of Explainable AI for mission-critical applications frequently found in, inter-alia, disaster monitoring and recovery, and emergency healthcare. We focus on scenarios where sequential critical decision making is made through evolving collaborative interactions between human and machine. Time-stamped feedback and forecasted feedback generated by human and machine will be leveraged to timely re-configure parameters of the underlying AI models embedded within the mission-critical applications. AI explainability will be examined using Information-theoretic tools to understand information transfer among external agents (human and machine) as well as across layers within the AI models in the machine in the presence of feedback. Depending on the common interests, selected case studies from disaster monitoring/recovery and medical emergencies will be used to inform problem formulation prior to methodology execution. This project requires understanding basic principles of machine learning and solid experience of software-based machine learning implementation (including but not limited to Python/Matlab). Knowledge and experience of Neural Network and Deep Learning architectures is not essential, but will be desirable.

Dr Derry Wijaya

Project Title: Extending Large Language Models (LLMs) capability for regional and low-resource languages

Project Description: LLMs (Large Language Models) have been making major breakthroughs in many natural language processing (NLP) tasks. However, LLMs are predominantly trained on and utilised for high-resource languages like English, Chinese, Spanish, French, which have abundant training data. The development in either pre-training, fine-tuning, or adaptation of LLMs and their utilisation for regional and low-resource languages, which are languages with few training data or available digital resources, often lack sufficient attention; even when these languages are spoken by millions of people around the world e.g., Javanese, Somali, Gujarati, Kazakh. This project will explore methods for extending the capability of existing LLMs for regional and low-resource languages through adaptations of these models to work with languages that have limited data availability. The project will also focus on evaluations of LLMs on these languages and their variations (code-mixing, informal forms) across a wide range of NLP tasks including ethical and bias evaluations. This project requires understanding of the principles of machine learning, neural networks, and deep learning architectures as well as a solid experience in their implementation frameworks (including but not limited to Python, Pytorch, HuggingFace, LLMs API such as OpenAI API, PaLM API) as well as running these ML workloads on any cloud.

Dr Arif Perdana

Project Title: Building Bridges: Safeguarding Ethical and Legal Boundaries in the Human-AI Coexistence Era

Project Description: This project delves into the critical area of responsible AI and seeks to explore the ethical, legal, and social implications of integrating AI into various aspects of our lives. This research aims to mitigate potential biases, privacy concerns, and societal impacts by exploring frameworks for responsible AI development. By exploring the dynamics of human-AI coexistence in multiple domains (education, finance, healthcare, and legal) this project seeks to develop policies and strategies to foster a harmonious relationship between humans and AI, where technology complements, enhances, and empowers human capabilities while upholding fundamental values and ensuring equitable outcomes. We expect potential candidates to have an academic or professional background in Information Systems, Law, or Computer Science.

Dr Risqi Saputra

Project Title: Spatio-Temporal Deep Learning for Semantic Segmentation of Mining Footprints using Multispectral Satellite Imagery

Project Description: In the past decades, there were significant attempts to map mining footprints on a global scale in order to better understand its impact to our environment and local communities. These include recognizing land use and land cover changes (LULC), identifying whether older mines are rehabilitated and restored. Oftentimes, manual land use identification using remote sensing data was performed to achieve those objectives. Those laborious manual tasks are still feasible for a local scale mapping but it will be too expensive to be replicated on a global scale. To this end, utilising autonomous mapping and classification systems such as based on deep learning algorithms are gaining more traction.

This project aims to develop novel deep learning algorithms capable of performing semantic segmentation of global mining footprints using multispectral satellite imagery. In particular, the project will invent a model that is capable of taking into account both spatial and temporal information embedded in the satellite images collected from multiple years. This is fundamental as some land use categories such deforestation can only be recognised only after identifying temporal changes on multiple images. In addition, the project might also explore super resolution techniques to increase the image quality of past satellite data as historical satellite images are usually only available in a very coarse resolution. This will require bringing together various state-of-the-art deep learning models in computer vision including but not limited to FPN, U-Net and its variant, GAN, and recent advances in processing sequential data (Transformers) or its varian (Vision Transformers).

Business Innovation

Dr Yessy Arnold Perangin Angin

Project title: The Implementation of Environmental, Social and Governance (ESG) Principles and its Impact

Project Description: This project investigates companies' implementation of ESG principles and their impact on performance metrics. While ESG principles have received attention for their adoption and effects on stakeholders, more is needed to know about their actual implementation and influence on the environment and company performance. The project analyses company practices and policies by examining sustainability reports, corporate governance structures, and environmental impact assessments. This evaluation provides insights into the level of commitment to sustainability and responsible governance.

Furthermore, the project assesses the relationship between ESG performance and overall company performance metrics. By analysing financial data, such as profitability, stock performance, and risk measures, alongside ESG ratings or scores, the study aims to uncover potential associations. Empirical analyses and statistical tests reveal the impact of ESG performance on financial outcomes, contributing to understanding the business case for ESG integration and long-term value creation.

For successful execution, a strong understanding of corporate finance and research methodology in finance is crucial, including familiarity with financial statements, valuation techniques, capital structure, and investment decision-making. Proficiency in research methodology is necessary to design a suitable framework, select relevant data sources, employ statistical techniques, and interpret results accurately.

Cybersecurity

Dr Rizka Purwanto

Project Title: Secure AI-Enhanced Light Curve Analysis for Satellite Detection in Space Surveillance

Project Description: In an era of increased space activities and satellite deployments, the need for effective space surveillance and monitoring is critical to ensure the safety and sustainability of space operations. To reduce the risk of space mission failures due to collisions, there have been efforts to collect and maintain knowledge of the space objects orbiting the Earth. Despite these efforts, there is still a lack of information on the size, shape, material, and orientation of space objects, which limits the accuracy of orbital predictions. To address this issue, a number of past works have studied various methods to perform space object identification. Photometric light curve analysis, which involves observing the changing brightness of satellites over time, offers a valuable method for identifying and tracking satellites, as it can be applied to small and dim objects across all orbital regimes. With the rapid emergence of research in deep learning, various models and architectures for performing sequence learning have been proposed in recent years, which showed promising results. However, there are still challenges in developing an accurate space object identification system due to the limited number of light curve dataset and the highly imbalanced datasets. Furthermore, it is also crucial to ensure that the Artificial Intelligence (AI) based framework is robust against adversarial attacks, during which adversaries craft malicious input to manipulate the result of the AI-based system.

This research project aims to develop a secure AI based approach for satellite detection using photometric light curve data to enhance space situational awareness. The proposed AI-driven framework has the potential to significantly contribute to the safety and sustainability of space operations. The successful implementation of this project will lead to a more comprehensive and efficient space traffic management system, which facilitates responsible space exploration and minimises the risks of collisions.

Dr Muhamad Erza Aminanto

Project Title: Enabling Privacy-Preserving Artificial Intelligence in Smart City Context

Project Description: Preserving privacy while deriving policy-relevant insights from citizen statistics poses a critical challenge for governments worldwide. This research integrates interdisciplinary perspectives to develop a holistic framework for privacy-preserving machine learning in the government context. Acknowledging the rise of personal data laws globally, we emphasise the importance of data privacy and explore the use of differential privacy as a technique to address privacy concerns during data analysis. However, recent studies indicate potential trade-offs in machine learning model performance when implementing differential privacy. Hence, our study aims to evaluate its impact on real data. We analyse the challenges and opportunities associated with privacy-preserving machine learning in government contexts. This research examines the delicate balance between privacy preservation, model accuracy, and policy formulation, considering factors from various aspects, as well as the broader societal implications of data-driven policies. This study will contribute to understanding the practicality and limitations of employing differential privacy in government settings, fostering interdisciplinary collaboration. Furthermore, this study should provide recommendations for organisations handling personal data, facilitating effective utilisation of privacy-preserving machine learning techniques. Ultimately, this interdisciplinary perspective empowers any organisation to make informed policy decisions based on accurate insights while safeguarding people's privacy.

Public Policy and Management

Dr Altaf Virani

Project Title: Governance and Policy Design Challenges of Public Sector Reform in Indonesia

Project Description: Indonesia is undergoing rapid development and societal change. Its public sector is at the heart of this transformation. The government has implemented big-ticket policy reforms across a range of sectors to improve public infrastructure, reduce poverty, provide universal health coverage, and tackle climate change and environmental sustainability. A crucial aspect of these reforms is how Indonesian policymakers are responding to Indonesia’s developmental challenge and addressing the policy dilemmas that come with changes in how public infrastructure is being created and how public services are being delivered.

This doctoral research project delves into the intricate governance and difficult policy design choices and challenges faced by Indonesia’s policymakers operating within a dynamic governance environment and the constraints and challenges that are typical of developing economies. The project seeks to develop a better understanding of how public policies can be designed to govern complex issues and effectively achieve critical policy goals in the face of contesting policy agendas, competing interests and pressures from diverse policy actors, changing policy priorities, and limited institutional capacities. By doing so, this research aims to contribute to emerging academic scholarship on policy design, advance theoretical work on new governance arrangements, and generate new empirical evidence and insights to inform the development and reconfiguration of policy solutions that Indonesian policymakers can draw upon.

Aspiring PhD students who are passionate about public policy research and possess an understanding of foundational public policy theories and concepts are encouraged to apply. Familiarity and interest in one or more policy areas, along with expertise in one or more qualitative or quantitative public policy research methods, are essential. Additionally, a desire to learn new research methods and demonstrate potential for producing high-quality research, as well as relevant practice or policy experience, will be advantageous.

Prospective candidates are encouraged to develop research proposals that align with its broad scope within their areas of familiarity and interest. They should outline their preliminary research ideas, the methodological approaches they intend to adopt, and the potential significance of their proposed work. Their research is expected to draw upon the expertise of Monash Indonesia academics from one or more disciplines, leverage transdisciplinary capacities, align with Monash's research priorities, and contribute to the advancement of public policy scholarship.

While the applicants have freedom to explore a broad range of developing ideas based on the project’s description and objectives, some possible research themes are suggested:

1. What mix of policy tools are effective in specific contexts and governance prerequisites for their success.
2. How policy entrepreneurs use narratives to drive policy change.
3. The role of policy capacity in effective policy design and good governance.
4. Policy design analysis to identify how design features impact policy outcomes.
5. Mechanisms to foster meaningful multi-stakeholder engagement in policy design to promote inclusive, responsive, and effective policies.
6. How policies are assessed, how evidence is used or misused in policymaking, and implications for policy and knowledge-translation.
7. How resilient and adaptive policy designs can better respond to policy crises and how crises can foster critical policy innovations.