I'm a data scientist and researcher specialized in natural language processing, multimodal learning, information visualization, and human-computer interaction. Using cutting-edge techniques from deep learning, machine learning, and statistics, I develop models to analyze and predict complex data.
You can find my resume here
Let's get in touch!
Email: fpeandrees[at]gmail.com
Building and training ML models for NLP tasks (e.g., topic modeling, hate speech detection) using NLTK, spaCy, Gensim, and HuggingFace.
Deep learning for vision and language using Pytorch. Focus on Transformers, VAEs, RNNs, GANs, LSTMs, CNNs, Difussion models.
Skilled in designing interactive visualizations with D3.js, Plotly, Seaborn, and Matplotlib using effective design principles and techniques.
Conducting user studies, prototyping and designing effective interfaces. Performing statistical analysis with Python/R.
Focus on Natural Language Processing, Multimodal Learning, and Information Visualization under the supervision of Dr. Giuseppe Carenini.
Relevant coursework: Multimodal Learning with Vision, Language and Sound (CPSC 532S), Commonsense Reasoning in Natural Language Processing (CPSC 532V), Discourse in NLP (CPSC 532G), Information Visualization (CPSC 547),
Computational Linguistics (CPSC 503), Topics in Human-Computer Interaction (CPSC 554).
Average grade 95%. Canadian GPA: 4.0/4.0
Focus on Social computing using Natural Language Processing, Deep Learning and Data Visualization techniques. Average grade 91/100. Canadian GPA: 4.0/4.0.
Focus on Information Retrieval, Software Engineering, and Project Management. Average grade 81/100. Canadian GPA: 3.7/4.0. Passed subjects: 66/66. (Best Graduated Student, Rank: 1/32)
Studied abroad at Politecnico di Milano in Milan, Italy from February-August 2016, gaining valuable experience and insights as a Computer Science student in an international setting.
Researched the use of large pre-trained models to detect hateful imagery. Developed a method to identify Antisemitic/Islamophobic social media posts. A dataset of Antisemitic/Islamophobic images is now available to aid researchers in creating better hate speech detection models. Supervisor: Dr. Savvas Zannettou.
Developed "TopicVisExplorer", a web-based tool that allows refinement and comparison of topic models with a novel topic similarity metric and refinement operations. A large user study was conducted to validate its usefulness. Supervisors: Dr. Evangelos Milios , Dr. Fernando Paulovich.
Developed a novel methodology for cross-language comparison of social media text to understand information privacy perspectives across different speakers. Analyzed unstructured textual social media data and published two conference papers on international computer science conferences. Supervisors: Supervisors: Dr. Cecilia Aragon (UW) , Dr. Claudia López (UTFSM).
Felipe González-Pizarro & Savvas Zannettou (To Appear). In International Conference on Web and Social Media (ICWSM), 11 pages. [Acceptance rate: 20%]
Raymond Li, Felipe González-Pizarro, Linzi Xing, Gabriel Murray & Giuseppe Carenini (Submitted). Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (ACL), 10 pages. [Acceptance rate: 31.4%]
Felipe González-Pizarro, Claudia López, Evangelos Milios, Fernando Paulovich & Marcelo Mendoza (Submitted). In ACM Transactions on Interactive Intelligent Systems, 30 pages. [Impact factor: 2.47]
Felipe González, Claudia López, Carlos Castro & Andrea Vasquez. (To appear). IEEE Latin America Transactions Journal, 7 pages. [Impact factor: 1.10]
Felipe González-Pizarro, Andrea Figueroa, Claudia López & Cecilia Aragon. In Computer Supported Cooperative Work (CSCW): The Journal of Collaborative Computing and Work Practices, 45 pages. [Impact factor: 1.912]
Felipe González, Andrea Figueroa, Claudia López & Cecilia Aragon. (2019, November). In Conference Companion Publication of the 2019 on Computer Supported Cooperative Work and Social Computing,CSCW'19, Austin, TX (2019), 4 pages. [Acceptance rate: 31.2%]
Felipe González, Yihan Yu, Andrea Figueroa, Claudia López & Cecilia Aragon. (2019, May). In Companion Proceedings of The 2019 World Wide Web Conference, WWW'19, San Francisco, CA(2019), 8 pages [Acceptance rate: 18%]
Felipe González, Claudia López & Carlos Castro. (2018, November). In 2018 37th International Conference of the Chilean Computer Science Society IEEE (SCCC'18), Santiago, Chile (2018), 8 pages. Best Paper Award.
Felipe González, Yihan Yu, Claudia López & Cecilia Aragon. (2018, November). Accepted on “Latin America as a Place for CSCW” Research Workshop of the 21st ACM Conference on Computer-Supported Cooperative Work and Social Computing
Python programming classes for freshman engineering students. Best ranked teacher for three semesters in a row. Class material (in Spanish) is available here
Data science classes for computer science & and engineering students. Developed a class curriculum, lesson plans, and instructions about how to manage data and create meaningful visualizations using Python, Pandas, Matplotlib, Seaborn and Plotly. Class material (in Spanish) is available here
I facilitate weekly workshops where students develop web/mobile app prototypes, conduct experiments/user studies, and receive constructive feedback from me. I also grade their reports to help them succeed. Supervisors: Dr. Izabelle Janzen
Teach students how to analyze and apply fundamental NLP algorithms and techniques. Text representation (e.g, language models), NLP Applications (E.g Topic modeling), Natural language understanding and generation Supervisors: Dr. Giuseppe Carenini, Dr. Vered Shwartz
Teach students about design and analysis of basic algorithms and data structures; algorithm analysis methods, searching and sorting algorithms, graphs and concurrency. Supervisor: Dr. Cinda Heeren
Promote management skills (planning, organization, direction and control) in students through the follow-up to their projects of the XXV Feria de Software (Software Fair). Compliance with requirements and design of prototypes are reviewed. Supervisors: Prof. Pedro Godoy, Prof. Sergio Murua
Teach students about the C/ C++ programming language. Teach students about data structures like linked list, stacks, queues, heaps, hash tables, trees and graphs. Supervisor: Dr. Diego Arroyuelo
Upload online resources and grade assignments & final projects. Supervisor: Katherine Rivera
Outstanding Graduate Student applicant. CAD 20,000 for living expenses.
ELAP Scholarship provides outstanding students from Latin America with short-term exchange opportunities for research in Canada at graduate levels. $7,300 for travel and living costs.
Got selected to attend to the CMMRS 2019 to learn about cutting-edge research in computer science at the Max Planck Institute for Software Systems, Germany. Travel and living costs fully funded.
Top 7% Applicant. $9500 for tuition and living expenses.
Got selected to attend to "La Serena School Data Science". Intensive week of interdisciplinary lectures focused on applied tools for handling big astronomical datasets. Travel and living costs fully funded. Our work project "Planetary Disk Direct Imaging" is available here
Travel Grant to support my participation at the 22nd ACM Conference on Computer-Supported Cooperative Work and Social Computing (CSCW 2019), Austin, TX, United States.
Travel Grant to support my participation at the 2019 International World Wide Web Conference, San Francisco, United States
$2600 for the research project: "Cultural Differences in Data Privacy Perspectives on Social Media".
Travel Grant to support my participation at the "Latin America as a Place for CSCW Research" workshop and at the conference on Computer-Supported Cooperative Work and Social Computing (CSCW 2018).
Travel Grant to support my participation at the 35th International Conference of the Chilean Computer Science Society (SCCC 2018)
Full tuition fees coverage on the MSc. Computer Science program
$6700 for living expenses
$5000 for travel and living expenses during a exchange experience at Politecnico di Milano, Milan, Italy.
Proposing a general-purpose visual commonsense generation model, VisualCOMET+, for the creation of structured commonsense triplets that can be used for downstream tasks. Also showing the success of the model in generating coherent and diverse topics for multimodal topic modeling.
Incorporating commonsense knowledge into CTM was explored using COMET and ConceptNet NumberBatch embeddings. Clustering was also investigated as an alternative to LDA-inspired techniques, with a discussion of the trade-off between the two and motivation to incorporate corpus-level semantic relationships into neural topic models.
A new topic modeling algorithm that incorporates commonsense knowledge to identify coherent topics in multimodal data is proposed. The algorithm was trained and evaluated on a dataset of 100K Antisemitic/Islamophobic posts from 4chan, demonstrating that injecting commonsense knowledge can improve the quality of topics generated by the algorithm.
Introducing an extended version of MultiModalTopicExplorer, a topic modeling visualization tool that allows users to identify and explore topics that align with their preferences and understanding of the domain. Results from a user study demonstrate improved task success rates and reduced physical and mental demands when using the user-adaptive version of the tool.
MultiModalTopicExplorer is an interactive visualization tool for topic modeling that addresses the limitations of current visualizations by allowing users to perform a qualitative analysis of results and displaying relevant images for each topic. A user study showed positive results in terms of task success.
