CCMBM/ci2 Advanced Machine Learning Workshop: Hands-on Experience for Image Classification
Date
This is a hands-on workshop focused on implementing and training an image classifier model to classify Covid-19 x-rays. We will showcase how to load and visualize data, balance and split your dataset, and train a model using the PyTorch framework.
Learning Objectives
-How to load data
-Data visualization
-Data splitting and balancing
-Loading a network
-Transfer learning
-Training Choosing loss function and architecture Hyperparameter optimization Overfitting
-Testing Error analysis Performance metrics
Participants will need to creat a Kaggle account https://www.kaggle.com/ prior to the workshop.
RSVP @ bit.ly/AdvMachLearning
Type
Time Duration
This is a hands-on workshop focused on implementing and training an image classifier model to classify Covid-19 x-rays. We will showcase how to load and visualize data, balance and split your dataset, and train a model using the PyTorch framework.
Learning Objectives
-How to load data
-Data visualization
-Data splitting and balancing
-Loading a network
-Transfer learning
-Training Choosing loss function and architecture Hyperparameter optimization Overfitting
-Testing Error analysis Performance metrics
Participants will need to creat a Kaggle account https://www.kaggle.com/ prior to the workshop.
RSVP @ bit.ly/AdvMachLearning
Speakers
Valentina Pedoia, PhD, is an Assistant Professor in the Radiology and Biomedical Imaging Department. She is an Imaging scientist with a primary interest in developing algorithms for advanced computer vision and machine learning to improve the usage of non-invasive imaging as diagnostic and prognostic tool. She obtained her doctoral degree in Computer Science working on features extraction from functional and structural brain MRI in subjects with glial tumors. After graduation, in 2013, she joined the Musculoskeletal and Imaging Research Group at UCSF as post-doctoral fellow to study degenerative joint diseases with compositional MRI techniques.
Dr Pedoia joined UCSF as Faculty In 2018. She is part of the Center of Intelligent Imaging where she serves as Co-director of the Educational Pillar. Her current research is focused in exploring the role of machine learning to extract imaging biomarkers of several musculoskeletal conditions including knee and hip Osteoarthritis, shoulder Instability and lower back pain. She develops analytics to model the complex interactions between morphological, biochemical and biomechanics aspects of the joints as a whole. Her goal is to develop efficient and effective data-driven models that able to extract imaging features and use them to identify risk factors, stratify patients and predict outcomes.
She has great interest in the clinical translation of novel technology, as such she is invested in making the image acquisition and processing faster, safer, and smoother for patients and clinicians.
I am a PhD candidate in the UC Berkeley-UCSF Joint Graduate Program in Bioengineering and mentored by Sharmila Majumdar, PhD, in the Department of Radiology and Biomedical Imaging at UCSF. My research focus is on the advancement of machine learning and computer vision techniques for medical imaging. More specifically, I aim to improve diagnostics of musculoskeletal disorders, such as low back pain and osteoarthritis, with magnetic resonance imaging by developing interpretable, quantitative analysis tools.
I am a graduate student in the UC-Berkeley/UCSF Graduate Program in Bioengineering with experience in medical imaging and data science. My work has centered around leveraging deep learning to accelerate acquisition of various MRI sequences, and in developing predictive models of disease progression. I am looking for future opportunities in MRI image reconstruction, and am also interested in applying my data science skills to other parts of the healthcare industry, such as pharmaceuticals and biotechnology.