From Economics to AI Research: Paulina Nemkova on PhD Life, Brain Reading, and Machine Learning

概览

This episode features Sam speaking with Paulina Nemkova, a second-year PhD student at the University of North Texas whose research focuses on artificial intelligence and machine learning. The conversation centers on how she moved from an economics background into AI research, what skills helped her make that transition, and what aspiring students should understand before pursuing academic work in the field.

A major theme is that AI and machine learning are open to people from varied backgrounds, especially when they bring quantitative foundations such as statistics, mathematics, or programming. Paulina emphasizes that academia requires a different mindset from industry: researchers must keep up with fast-moving literature, contribute new knowledge, and be comfortable spending long stretches reading, thinking, and working independently.

The episode also introduces Paulina’s current research on EEG-based “brain reading,” where machine learning models classify brain-signal data to infer the category of objects a person may be thinking about. The discussion connects this work to possible medical, forensic, and psychiatric applications while noting that true mind reading or modeling consciousness remains a long-term challenge.

分段落总结

[00:01] Episode Introduction

[事实] Sam opens the third coffee chat session on his channel, Data Science with Sam.

[事实] He introduces Paulina Nemkova as a guest from academia who will discuss her learning and research journey in artificial intelligence and machine learning.

[00:28] Paulina’s Background

[事实] Paulina says she is a second-year PhD student at the University of North Texas.

[事实] Her research area is AI and machine learning.

[事实] She says she can discuss how to enter a PhD program in AI and what students should know before choosing that path.

[01:24] Why She Chose AI Research

[事实] Paulina says her path was not straightforward, because her bachelor’s degree was in economics from Belarus.

[事实] After moving to the United States, she decided she needed graduate education and considered multiple fields.

[事实] She wanted a field that would be valuable in the future, exciting to her, and applicable across many areas.

[事实] AI felt like a strong fit because she already knew statistics and could apply AI to many different interests.

[02:45] Programming and Preparation Before the PhD

[事实] Paulina says she had learned C++, Turbo Pascal, Python, MATLAB, and R before entering the program.

[事实] She also took classes before applying for the PhD.

[事实] Before applying, she spent about a year contacting professors and doing research with some of them.

[事实] She contacted around 50 professors; about 10 replied, and about five suggested working with her.

[推测] Her route suggests that direct outreach and early research exposure can help applicants test whether their skills are sufficient for a research area.

[04:10] Changing Fields Into AI

[事实] Sam says Paulina’s story shows that a person can enter AI or machine learning from a different background if they have some programming, statistics, or mathematical skills.

[事实] He frames her transition from economics into AI research as useful encouragement for aspiring data science and machine learning learners.

[05:16] Academic Mindset and Competition

[事实] Paulina says she is speaking mainly from an academic perspective.

[事实] She describes academia as competitive because researchers are judged by whether they can discover something new.

[事实] She contrasts this with industry, where people may repeatedly apply existing knowledge.

[事实] She says academic expectations are high because researchers need to produce new work.

[06:11] Staying Current in a Fast-Moving Field

[事实] Paulina says AI is developing so quickly that work published or discovered a few years ago may already be outdated.

[事实] She says researchers must stay aware of recent papers, news, and lab developments to remain relevant.

[推测] The episode presents continuous literature tracking as a core survival skill for AI researchers, not just an optional habit.

[07:06] Academia Versus Industry

[事实] Paulina says she knew for a long time that she preferred academia, theory, and discovery over repeatedly applying knowledge she already had.

[事实] She says aspiring academics should know that research involves a lot of reading, thinking, and working alone with a laptop.

[事实] She notes that conferences may look exciting, but much of the actual work is solitary.

[事实] She says starting her degree during COVID-19 added to that experience of working alone.

[08:25] Rapid Growth of AI Research

[事实] Sam gives an example from his own deep learning work on COVID-19 detection from chest X-ray images.

[事实] He says that while writing a paper, he found many related works had already appeared within about six months.

[事实] He mentions Connected Papers as a site that helps show relationships among papers in a research area.

[事实] He says convolutional neural networks and deep learning are still evolving and often involve trial-and-error approaches.

[09:50] Why the Field Remains Exciting

[事实] Paulina agrees that much remains undiscovered in AI.

[事实] She compares current discoveries to a tiny visible part of a much larger field.

[事实] She says this uncertainty and room for discovery are what make AI exciting to her.

[10:26] Exploring AI Subfields Through Projects

[事实] Paulina says that as a second-year PhD student, she is still taking classes alongside research.

[事实] She says PhD classes can lead to projects driven by curiosity.

[事实] She has worked on projects in NLP, deep learning, regular machine learning, and varied applications.

[事实] She recommends exploring different AI areas before committing to one.

[事实] She says her projects are usually uploaded to GitHub and are open for others to view.

[11:27] Main PhD Research: EEG-Based Brain Reading

[事实] Paulina’s main PhD project focuses on what she calls “brain reading.”

[事实] The project uses a large dataset of EEG signals from the brain.

[事实] The team is doing classification to understand what a person may be thinking about.

[事实] The original motivation was to help people with locked-in syndrome, who can think but cannot express themselves because their bodies are paralyzed.

[推测] The project is positioned as assistive communication research rather than literal full mind reading.

[12:35] Replication and Research Integrity

[事实] Paulina says the project began by replicating similar work from Stanford University.

[事实] Her team aims to replicate, improve, and build on that prior work.

[事实] She says they are in touch with Stanford professors.

[事实] She emphasizes that replication is important in academia because research that cannot be replicated creates problems.

[事实] She also notes that statistics can be misused and results can be reported in different ways.

[13:28] Cryptocurrency Time Series Research

[事实] Paulina says she has also published a paper on time series analysis of cryptocurrency prices.

[事实] She says this work used part of her economics background.

[事实] She is also interested in cryptocurrencies and decentralized finance.

[14:10] Industry and Academia Collaboration

[事实] Sam says his finance background makes him interested in Paulina’s cryptocurrency work.

[事实] He suggests that future collaboration could happen around deep learning and related projects.

[事实] He says industry and academia can benefit from working together, with industry offering expertise that may help academic researchers.

[15:22] Connecting Computer Science and Business

[事实] Paulina says she started a startup club at her school.

[事实] She says she noticed there was not enough collaboration between the computer science department and the business school.

[事实] Her goal was to create more meetings and conversation between those groups.

[事实] Sam says he will share Paulina’s GitHub link and encourages viewers to contact her if they are interested in her research or collaboration.

[16:16] Future Applications of Brain-Signal Models

[事实] Sam asks how Paulina sees the future of AI evolving around her neuroscience-related research.

[事实] Paulina says a strong model for this task could benefit many fields.

[事实] She names medicine, forensics, and psychiatry as possible application areas.

[事实] She says she will not go too far in speculation because the brain is still not well researched.

[18:09] Limits of Current Brain Reading

[事实] Paulina clarifies that her research is not exactly predicting thoughts.

[事实] The current task predicts the class of object that a person may be thinking about.

[事实] She says this is a step toward mind reading, but there is still a long way to go.

[事实] She believes deep learning and AI may make progress in this area faster and more reasonable in the future.

[18:44] AI, Neuroscience, and Long-Term Potential

[事实] Sam mentions Neuralink and work involving a monkey playing a ping-pong game using brain signals.

[事实] He says many efforts are happening around applying AI and machine learning to neuroscience.

[事实] He says AI and machine learning may help answer long-standing scientific questions, but the field is still taking early steps.

[推测] Sam views Paulina’s research area as potentially revolutionary if it can overcome current scientific and technical limits.

[20:11] Advice for Aspiring Students

[事实] Sam advises viewers not to feel limited by their original background if they want to pursue AI and machine learning research.

[事实] He says people can make a fresh start at any time.

[事实] He points back to Paulina’s journey as an example of changing direction through mindset and preparation.

[21:03] Paulina’s Closing Invitation

[事实] Paulina thanks Sam for having her on the session.

[事实] She invites people interested in academia to reach out to her for advice.

[事实] She mentions her website, neumkowa.com.

[事实] She says she wants to help others because many people helped her.

[事实] She says getting into school is its own project, and knowing the steps can make the process more efficient.

[22:04] Final Closing

[事实] Sam says more people from academia should help aspiring AI and machine learning students.

[事实] He says AI and machine learning still have many things to learn and that the learning curve does not end.

[事实] He thanks Paulina for joining and says future coffee chat sessions will feature other industry or academic experts.

播客点评/总结

This episode is valuable as a practical, beginner-friendly discussion of entering AI research from a nontraditional background. Its strongest parts are Paulina’s concrete account of contacting professors, building research exposure before applying, and learning how academic AI work differs from industry practice.

The discussion also gives listeners a useful view of how broad AI research can be, from NLP and deep learning projects to EEG-based brain-signal classification and cryptocurrency time series analysis. The episode avoids presenting AI as a single narrow career track and instead frames it as a flexible field connected to economics, neuroscience, medicine, finance, and entrepreneurship.

[推测] The main limitation is that the episode stays at a high level and does not go deeply into technical methods, datasets, model architectures, or PhD application details. It is best suited for students and early-career learners exploring whether AI research or graduate study might fit them, rather than listeners seeking a technical tutorial.