Farhad Pourkamali Anaraki
Assistant Professor, Department of Mathematical and Statistical Sciences
Director of Mathematics, Information, and Data Science (MINDS) Lab
University of Colorado Denver
Farhad Pourkamali is currently a tenure-track Assistant Professor
in the Department of Mathematical and Statistical Sciences at
the University of Colorado Denver (CU Denver). He earned his Ph.D. in Electrical
Engineering from the University of Colorado Boulder (CU Boulder), where he subsequently
completed a one-year postdoctoral fellowship in the Department of Applied
Mathematics. Prior to his current role, he served as an Assistant Professor
in the Computer Science Department at the University of Massachusetts
Lowell (UMass Lowell) from 2018 to 2022. In 2021, he received the Visiting Faculty Research Program (VFRP) award
from the Air Force Research Laboratory Information Directorate (AFRL/RI). Since the fall of 2022, he has been a part
of CU Denver, where he leads the Mathematics, Information, and Data Science
(MINDS) Lab. His main emphasis is on advancing machine learning techniques to
reduce the necessity for human input, enhance their computational and statistical efficiency,
and create accessible cyberinfrastructure resources. He is committed to addressing practical problems
in various domains, including materials science and natural hazards analysis, and is actively seeking
opportunities for collaboration.
The objective of my research is to develop learning algorithms
that work for dynamic and multi-modal data with minimal human intervention.
Dynamic Data: Addressing dynamic data is crucial as
the world is full of evolving, non-static information. Algorithms
capable of adapting to changes over time without manual updates
are essential for real-time applications, such as advanced manufacturing,
weather forecasting, and precision medicine.
Multi-Modal Data: Multi-modal data refers to information that
comes in various forms, such as text, images, audio, and video.
Learning algorithms that can effectively integrate and
interpret this diverse data can lead to more robust and
versatile applications. This is particularly relevant in fields
like autonomous driving, where the system must understand visual,
auditory, and sensor data, or in healthcare, where patient data might
include clinical notes, radiology images, and lab results.
Minimal Human Intervention: The emphasis on reducing human intervention is
crucial for scalability and accessibility. Algorithms that require minimal fine-tuning
and human guidance can be more easily deployed across different domains
and by users with varying levels of expertise. This is particularly
important for applications in resource-limited settings or where
rapid deployment is necessary.
Interdisciplinary Collaboration: I always look for interdisciplinary collaborations, as developing algorithms for dynamic and multi-modal data often
requires expertise from computer science, statistics, cognitive science, engineering, and domain-specific knowledge.
The complete collection of my publications can be found on Google Scholar.
- Evaluation of Classification Models in Limited Data Scenarios with Application to Additive Manufacturing
- This paper presents a novel framework that enables the generation of unbiased estimates for test loss using fewer labeled samples,
effectively evaluating the predictive performance of classification models in data-limited applications. The framework's key innovation
lies in developing an adaptive sampling distribution that iteratively identifies
influential testing samples based on interactions between learner and evaluator agents.
Notably, the adaptive distribution dynamically adjusts the evaluator agent's
supervisory role by prioritizing inputs with discrepancies between the agents
and considering the evaluator's uncertainty. Comprehensive experimental
analyses on synthetic data and two sparse data sets from material extrusion
additive manufacturing problems validate the framework's superiority over
uniform and fixed sampling distributions. First, the proposed framework
provides unbiased estimates of the test loss across various data sets,
sampling ratios, and evaluator models. Second, the introduced adaptive
sampling distribution significantly reduces the standard deviation of the
test loss estimator compared to uniform sampling, achieving a 50% reduction
for a 10% sampling ratio in the filament selection benchmark. Third,
the framework demonstrates its efficacy in model selection to determine
the optimal number of hidden units with a reduced number of test samples.
Overall, this work offers a promising framework for evaluating classification
models in applications where acquiring labeled data is time-consuming and
resource-intensive, including materials science and engineering.
- Two-Stage Surrogate Modeling for Data-Driven Design Optimization with Application to Composite Microstructure Generation
- This paper introduces a novel two-stage machine learning-based surrogate modeling
framework to address inverse problems in scientific and engineering fields. In the first stage of the proposed framework,
a machine learning model termed the "learner" identifies a limited set of candidates within the input design
space whose predicted outputs closely align with desired outcomes. Subsequently, in the second stage,
a separate surrogate model, functioning as an "evaluator," is employed to assess the
reduced candidate space generated in the first stage. This evaluation process eliminates inaccurate
and uncertain solutions, guided by a user-defined coverage level. The framework's distinctive
contribution is the integration of conformal inference, providing a versatile and efficient
approach that can be widely applicable. To demonstrate the effectiveness of the proposed framework
compared to conventional single-stage inverse problems, we conduct several benchmark tests and investigate an
engineering application focused on the micromechanical modeling of fiber-reinforced composites. The results
affirm the superiority of our proposed framework, as it consistently produces more reliable solutions.
Therefore, the introduced framework offers a unique perspective on fostering interactions between machine
learning-based surrogate models in real-world applications.
- High-Throughput Design and Analysis of Novel Ceramics
- Funding Agency: Army Research Lab (ARL)
- Data Science Approaches to Advance High Solids Loading Additive Manufacturing
- Funding Agency: Army Research Lab (ARL)
- Multi-Scale Models based on Machine Learning and a Fiber Network Model
- Funding Agency: National Aeronautics and Space Administration (NASA)
- Learning from Imbalanced Data with Confidence and Minimal Supervision
- Funding Agency: Air Force Research Laboratory Information Directorate (AFRL/RI)
- MATH 6388 Statistical and Machine Learning, Fall 2022 and Fall 2023
- MATH 4/5388 Machine Learning Methods, Spring 2023
- MATH 1376: Programming for Data Science, Spring 2024
- Matthew Knodell (Applied Math PhD Student, 2024-present)
- Courtney Franzen (Applied Math PhD Student, 2024-present)
- Rachel Drummond (Applied Math PhD Student, 2023-present)
- Colin Furey (Applied Math PhD Student, 2023-present)