My Story
Rick Rejeleene
http://rickrejeleene.me/
Texarkana, Texas: 2021
Education & Career
Currently: Ph.D. in Machine Learning at U of Arkansas
Dr.Xiaowei Xu, inventor of Dbscan (PhD Machine Learning, Germany)
Dr.Jae Hyun Kim Ph.D. in Bioengineering (Kansas), PhD Computer Science -Korea
Previously:
M.S in Computer Science - University of Rhode Island, Kingston, Rhode Island, U.S.
Thesis: Kodai - Segmentation Software, Dr. Joan Peckham, former Dept. Chair of Computer Science
Data Scientist: Cru, non-profit, Orlando, Florida
Product Manager: Triblio (USD - 1.2 Million to 3.2 Million), Washington D.C.
Hobbies: 40+ Poetry, Making friends & Meeting Internationals being curious about their culture
Fun Fact: Books: 504 read, 83 books this year [Computing, History, Psychology, Sociology, Religion]
Last book read: Christopher Bishop’s pattern recognition work
What others say of me: Approachable, Loyal, Visionary, Listener, Tenacious, Intellectual
Research summary
Kodai: Software Segmentation
Idea: Software architecture for segmentation models
Issue: Big data
Tools: Django, Python, ElasticSearch
MovieBarCode: Summarize Video into Bar-Code
Idea: Summarize Video into BarCode
Issue: Compress large-scale videos
Tools: OpenCV, Python, Front-end tools
Research summary
Title: “Kodai: Towards Creativity in Machine Learning”
Area: Neuroscience, Machine Learning & NLP
Big Vision: Creativity in Human Brain & Language Models
Why? “Creativity is what sets us humans apart”
Definition: Surprise, Novel, Useful [Margaret Boden]
Background:
1. Anatomy of Brain
2. Neuroscience (Cognitive Models) - Anna Abraham, Rex Jung
3. Language Models (Transformer, BERT, GPT-2)
Research summary
Where is Creativity Generated in Human Brain?
Frontal Lobe, Combination of (a) & (b) & (c )
a. Default Mode Network
b. Salience Network
c. Executive Network
GAPS: Quantifying representing them mathematically
We propose:
1. Creative Function:
A novel machine learning score for describing creativity in text
2. Creative Oriented Architecture:
A novel creative machine learning framework, for generating coherent text
3. Controlled Text Generation:
A novel machine learning based parameter for giving control over output attributes
Research summary
Kodai: Towards Creativity in Machine Learning
Research Direction:
1. How to replicate creativity quantifiably?
2. How to build creative language processing of brain?
3. Compare language processing of brain & existing NLP systems
4. Why are Language models not able to generate coherent creative text?
Engineering:
5. Demonstrate Creativity
6. Compare Nobel Prize Winner vs Normal Writings
7. Use creative score to distinguish between Writings
a. Similar to Economic complexity
Planned Outputs of work:
b. We propose a Creativity Index
8. Publication (5-6) (ML, NS, CogSci) c. Using it, we score of creativity in Language using LLM
9. Book (Introduction to Computation Models in Neuroscience) d. Using it, we can create map of writings in literature
Social Impact of our work:
10. Creative Assistant Writing - Able to write better through generation of accurate prompts
11. Creative Speech Assistant - Able to respond through generation of accurate speech prompts (Alexa, Google Home)
12. Generate Creative Artifacts - Able to generate text, video, audio
