Arya Karami

Arya Karami

Researcher of Machine Learning

University of New South Wales

Biography

Greetings! I am an experienced data scientist specializing in trans-disciplinary statistical and deep learning, currently affiliated with the University of New South Wales (UNSW). My expertise encompasses Statistical Network Modeling, Stochastic Processes, and the innovative application of artificial intelligence in emerging industries.

Explore my portfolio, featuring projects that highlight my proficiency in navigating the intersection of statistical modeling and AI applications for the advancement of various industries. I am passionate about using cutting-edge technologies to make a meaningful impact.

In my spare time, I dedicate myself to studying philosophy, physics, and running.

Interests
  • Statistical Network Modeling, Social Network Analysis.
  • Bayesian Statistics, Neural Networks.
  • Spatiotemporal Modeling and Timeseries.
  • Stochastic Processes, Stochastic Programming, and Dynamic Programming.
Education

  • RA in Statistics

    Univeristy of New South Wales

  • RA in Industrail Engineering

    Sharif University of Technology

  • MEng in Industrial Engineering (Systems Optimisation)

    Sharif University of Technology

  • BSc in Industrial Engineering

    Sharif University of Technology

Projects

*
All NeST Stochastic Processes SNA
Network Stochastic Processes and Time Series (NeST)
In this research, we closely explore the Hawkes Network Stochastic Processes and Time Series (NeST), a method for understanding how events depend on each other. Over the past 50 years, we’ve seen different versions of this process. We aim to create a single, easy-to-use solution for 12 variations. By doing this, we hope to make it simpler for everyone to study and predict how things relate in various fields, all while building on the decades-long history of Hawkes models. We introduce a practical solution by using a general Bayesian approach.
Network Stochastic Processes and Time Series (NeST)
A Novel Algorithm for Information Cascade Prediction using Point Processes
This Project introduces a pioneering algorithm for predicting information cascades in dynamic social networks. The application of point processes for information cascade prediction in social networks involves modeling the temporal dynamics of information propagation. Point processes, such as Hawkes processes, capture the self-exciting nature of events and inter-event dependencies. By analyzing past cascades, these models predict future information diffusion patterns, aiding in understanding and managing viral content spread in dynamic social network environments. This approach contributes valuable insights to fields like marketing, epidemiology, and online influence dynamics.
A Novel Algorithm for Information Cascade Prediction using Point Processes
Developing ERGM models for Social Networks
Developing Exponentially Random Graph Models for Social Network Analysis. Exponential Random Graph Models (ERGM) serve as a powerful tool for Social Network Analysis (SNA). These models are employed to understand and interpret the complex patterns of relationships within social networks. ERGMs go beyond simple descriptive statistics, allowing researchers to statistically model the likelihood of observed network structures.
Developing ERGM models for Social Networks

Recent Talks

Social Network Monitoring
A seminar at Sharif University of Technology’s Industrial Engineering Department in honor of Research Week.
Dec 20, 2023 1:00 PM — 3:00 PM Online seminar
Social Network Monitoring
A Novel Latent Variable Approach to Bayesian Inference for Hawkes Processes
This study is dedicated to exploring the latent variable approach for Bayesian inference in the context of parameter estimation for Hawkes processes. By incorporating latent variables within the Bayesian framework, the research aims to refine the process of parameter inference. Hawkes processes, frequently employed for modeling self-exciting events, stand to benefit from this approach, potentially enhancing the effectiveness of parameter estimation. The study contributes valuable insights into the application of Bayesian methods for improved parameter inference in the domain of temporal point process modeling.
Dec 10, 2023 — Dec 15, 2023 University of Wollongong
A Novel Latent Variable Approach to Bayesian Inference for Hawkes Processes
Exploring Organizational Social Capital in Dynamic Ego Networks
In this presentation, we address challenges in measuring organisational social capital, crucial for coordination and cooperation. we explore advancements in social network analysis and data analytics to overcome biases in subjective methods. The proposed approach uilizes novel statistics for continuous monitoring, demonestrated efficiently with real-wrold datasets. By refining social capital measurement, it provides deeper insights into organisational dynamics, informing targeted interventions for social capital development within organizations.
Sep 12, 2023 1:00 PM — Sep 13, 2023 3:00 PM Hybrid Conference
Exploring Organizational Social Capital in Dynamic Ego Networks

Recent Publications

Arya Karami (2023). An Online Support Vector Machine Algorithm for Dynamic Social Network Monitoring. Journal of Neural Networks.

PDF Cite DOI

Arya Karami (2020). The multi-objective supplier selection problem with fuzzy parameters and solving the order allocation problem with coverage. Journal of Modelling in Management.

PDF Cite DOI