Our team leads the Explainable Universe working group at the CosmicAI Institute, the NSF-Simons AI Institute housed at the University of Texas at Austin. CosmicAI aims to grow transformative AI advances, reform research workflows, and increase astronomy and AI accessibility through developments in four fundamental AI pillars: trustworthiness, robustness, explainability, and efficiency. The Explainable Universe working group aims to extend explainable and interpretable AI methods and unify them with causal reasoning, enabling explanation of the underlying mathematical relationships, while remaining robust to noise and uncertainty.
Collaborate with Us: We welcome practitioners in academia and industry to connect with us to explore synergies and demonstrate the broader impact of the tools we are developing across various fields. If you're interested in discussing potential collaborations or partnerships, reach out to us -- we'd love to chat! Email:arya.farahi@austin.utexas.edu
Astrostatistics
We develop and use advanced statistical and machine learning methods to unlock the mysteries of the universe. Through the application of cutting-edge statistical techniques and models to complex astronomical data, we strive to provide new insights and understanding of cosmic phenomena.
Our goal is to accelerate to the advancement of astronomy and astrophysics by developing innovative statistical tools and collaborating with researchers across different fields. By combining expertise in statistics and astronomy, we seek to push the boundaries of knowledge and make significant contributions to our understanding of the cosmos. and fundamental physics.
Projects
Detection of anti-correlation of hot and cold baryons in galaxy clusters
A Machine Learning Approach to Enhancing eROSITA Observations
Percent-Level Test of Isotropic Expansion Using Type Ia Supernovae
PoPE: A population-based approach to model spatial structure of astronomical systems
Approximate Bayesian Uncertainties on Deep Learning Dynamical Mass Estimates of Galaxy Clusters
Gaia DR2 unravels incompleteness of nearby cluster population: New open clusters in the direction of Perseus
Trustworthy AI for Societal Changes
Our mission is to advance fairness, transparency, and trust in AI systems. We achieve this by developing and evaluating algorithms to ensure their impartiality and reliability. Our aim is to combat algorithmic bias and discrimination in critical areas, including climate, housing, healthcare, and urban planning.
Through collaboration with diverse stakeholders, including city government, non-profits, and private sector, we advocate for the ethical use of AI systems and the development of transparent and equitable algorithms. Our commitment to advancing research in algorithmic fairness and trustworthiness drives us to make a positive impact on society.
Projects
A Social Justice Framework for Urban Communities
Modeling and Quantifying Group Disparities
A Case Study of Homeownership Racial Disparity
ActiveRemediation: The Search for Lead Pipes in Flint, Michigan
KiTE - An auditing tool for trustworthiness hypothesis testing and quantification
TAME Pain: Trustworthy AssessMEnt of Pain from Speech and Audio for the Empowerment of Patients
Auditing AI Systems
Our team is at the forefront of ensuring AI safety and integrity, specializing in the development and deployment of advanced auditing systems. We work closely with third parties, providing expert consultation to guide the implementation of these systems effectively. Our focus is on rigorously evaluating AI technologies for performance, fairness, and trustworthiness, ensuring they meet strict criteria for safety and ethical standards. By assisting in the deployment of comprehensive auditing measures, we help maintain the reliability of AI systems, ensuring they operate not only efficiently but also in a manner that is equitable and safe for all users.
This multifaceted approach involves rigorous testing, analysis, and refinement to identify and mitigate biases, enhance transparency, and improve the overall reliability and ethical standards of AI technologies. Through our work, we are committed to fostering an AI ecosystem that is both powerful and responsible.
We collaborate with
CosmicAI
The CosmicAI Institute grows transformative AI to meet Astronomical challenges through research in four fundamental AI themes: trustworthiness, efficiency, interpretability, and robustness. AI and astronomy experts co-lead each research challenge.
A grand challenge at the University of Texas at Ausitn that builds AI-based technologies helping us solve complex problems in nearly every discipline and industry.
Rubin Observatory is an observatiory that conducts a ten-year survey of the Southern Hemisphere sky with the goal of answering some of scientists' biggest questions about the Universe.
The DREAMS project uses large suites of comprehensive galaxy formation simulations to study the nature of Dark Matter. DREAMS utilizes Machine Learning to isolate the impact of Dark Matter and discover the fundamental properties of Dark Matter with astrophysical observables.
HEAD mission is to form an international consortium at the nexus of AI, healthcare, and ethics, fostering a collaborative environment to address these critical challenges. Our vision is to reshape policy and infrastructure to make technology-driven healthcare universally accessible and equitable.
The Cosmostatistics Initiative (COIN) is a worldwide endeavor aimed to create an interdisciplinary community around data-driven problems in Astronomy. It is designed to promote innovation in all aspects of academic scientific research.
