Work Experience
- Lead the development of an asynchronous job submission workflow that delivers optimal microscopy sampling locations for a single page application that displays integrated circuit layout clips.
- Orchestrated the compartmentalization, organization and packaging of 4 Python / JavaScript mixed codebases.
- Architected and lead the development of an internal single page application for clustering and visualizing tens of millions of integrated circuit layout clips for comparing products. Responsibilities included devising and implementing data generation algorithms, data automation, UI/UX design, full-stack development, benchmarking, parallelization and performance improvements, maintaining nightly builds.
- Initiated and managed the adoption of developer productivity and code quality improvement tools for the whole project team, such as TypeScript, Vite, swc.
- Improved the robustness of deep neural networks against adversarial attacks and noise using sparse encoding, dictionary learning, quantization, population coding, and Hebbian learning.
- TA for 6 courses in a wide range of fields. As the head TA, designed and implemented the course project for the Data Science Capstone course.
- Grouped and reduced Optane die-level failure maps to fundamental failure modes using dimensionality reduction, clustering and sparse dictionary learning. Visualized the results using a Dash/Plotly based prototype UI.
- Built signal processing tools for the analysis of vocalization, video and neural implant recordings of mice and rats from experiments, in MATLAB and Python.
- Converted Visual Basic driver code to C using libusb-hidapi for a modernization of medical gamma spectrum analyzer project.
- Built a local web server running on a single board computer using LAMP stack and jQuery to serve data coming from the spectrum analyzer in real time.
Education
- Lab: Wireless Communications and Sensornets Laboratory
- Advisor: Upamanyu Madhow
Programming Experience
Awards and Honors
Publications
Can Bakiskan, Metehan Cekic, Upamanyu Madhow, "Early Layers Are More Important For Adversarial Robustness," in International Conference on Machine Learning (ICML) Workshop - New Frontiers in Adversarial Machine Learning, 2022
Metehan Cekic, Can Bakiskan, Upamanyu Madhow, "Layerwise Hebbian/anti-Hebbian (HaH) Learning In Deep Networks: A Neuro-inspired Approach To Robustness," in International Conference on Machine Learning (ICML) Workshop - New Frontiers in Adversarial Machine Learning, 2022
Can Bakiskan*, Metehan Cekic*, Upamanyu Madhow, "Neuro-Inspired Deep Neural Networks with Sparse, Strong Activations," in IEEE International Conference on Image Processing (ICIP), 2022
Metehan Cekic, Can Bakiskan, Upamanyu Madhow, "Towards Robust, Interpretable Neural Networks via Hebbian/anti-Hebbian Learning: A Software Framework for Training with Feature-Based Costs," in Software Impacts Journal, 2022
Can Bakiskan, Metehan Cekic, Ahmet Dundar Sezer, Upamanyu Madhow, "Sparse Coding Frontend for Robust Neural Networks," in International Conference on Learning Representations (ICLR) Workshop on Security and Safety in Machine Learning Systems, 2021
Can Bakiskan, Metehan Cekic, Ahmet Dundar Sezer, Upamanyu Madhow, "A Neuro-Inspired Autoencoding Defense Against Adversarial Attacks," in IEEE International Conference on Image Processing (ICIP), 2021
Can Bakiskan, Soorya Gopalakrishnan, Metehan Cekic, Upamanyu Madhow, Ramtin Pedarsani, "Polarizing Front Ends for Robust CNNs," in IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), 2020
*: Equal contribution
Teaching Experience
Data Science Capstone
Communication Systems Design
Digital Communication Fundamentals
Machine Learning from Signal Processing Perspective
Signal Analysis
Introduction to Fields and Waves