Sushma Anand Akoju

My past 10 years of work experience, research focus and achievements: tinyurl.com/intro-sushma-anand-akoju

This sets a precedent for my research goals and serves as documentation should anyone attempt targeted harassment or coercion or IPV again.

My learning from different mentors draws the analogy to Princess Ida in Monument Valley -knowledge transferred as torches/clues between mentors and students. I maintain open, safe conversations with students for mutual learning.This reasserts that intellectual domination is only an illusion.

#learning-as-a-hobby #naturallanguageunderstanding #neurosymbolic

research-experience/learning-as-a-hobby/neurosymbolic-methods-study.md at main · sushmaakoju/research-experience Contribute to sushmaakoju/research-experience development by creating an account on GitHub.

(7) Symbolic encodings in Deep Learning . These seven domains as 360-degree approach enables breakthrough insights at the intersections of Neurosymbolic methods and Natural Language Understanding as a research topic. tinyurl.com/neurosymboli...

My study and research on Natural Language Understanding spans 7 integrated domains: (1) Deep Learning architectures (2) Theoretical Logic (3) Automata, Grammars & Theory of Computation (4) Neuroscience of Language (5) Cognitive Science approaches to AI (6) Neuroscience of reasoning, and

research-experience/learning-as-a-hobby/neurosymbolic-methods-study.md at main · sushmaakoju/research-experience Contribute to sushmaakoju/research-experience development by creating an account on GitHub.

Deep learning meets symbolic reasoning through sustained my self-directed study across 50+ topics. tinyurl.com/neurosymboli...

research-experience/learning-as-a-hobby/neurosymbolic-methods-study.md at main · sushmaakoju/research-experience Contribute to sushmaakoju/research-experience development by creating an account on GitHub.

My neurosymbolic methods journey spans 8+ years: from automated theorem proving & first-order logic (2018) to Natural Logic (2022) & AMR parsing (2021), plus hands-on implementation with Z3 solvers (2022), polynomial parsers (2020), and MLN applications (2021). tinyurl.com/neurosymboli...

research-experience/carnegie-mellon-university/README.md at main · sushmaakoju/research-experience Contribute to sushmaakoju/research-experience development by creating an account on GitHub.

Technical expertise across NLU, multimodal perception, and neurosymbolic reasoning. tinyurl.com/cmu-experience

My 5+ years at Carnegie Mellon spanned multiple cutting-edge projects: Conversational AI with SARA (Socially-Aware Robot Assistant) under Profs. Tom Mitchell & Justine Cassell, hierarchical ML under Prof. Barnabas Poczos, plus HuBMAP computational biology & robotics work. tinyurl.com/cmu-experience

Authorship clarification: For any authorship verifications/if you hear someone else taking credit for my research: contact sakoju2014@gmail.com.

Similar symbolic encoding spaces enable effective human-AI collaboration, but this similarity should not be interpreted as evidence that LLMs 'understand' or 'reason' in human terms. The functional similarity ≠ cognitive equivalence, as experts consistently note.

This makes them natural collaborators in expression for transforming communication: t.co/t6WlGabdbj

A public summary describing details of these relevant encoding spaces between humans and AI Assistants/LLMs is here: tinyurl.com/all-about-la...
#AIforGood #BetterConversations

Recent research on language as symbolic communication (auditory, visual, written) reveals why LLMs are effective collaboration tools: they operate in the similar symbolic encoding spaces that encode thoughts, beliefs, multimodal perceptions & experiences. #academicversion tinyurl.com/all-about-la...

research-experience/logical-fallacies-case-study.md at main · sushmaakoju/research-experience Contribute to sushmaakoju/research-experience development by creating an account on GitHub.

tinyurl.com/logical-fall... But when false clsims enter legal terrain, it can set logical traps such as liar’s paradoxes so either admitting or denying the allegation of making false allegations, after contrary evidence emerges, leads to potentially problematic legal outcomes.

As I have a Masters in Forensic Science and criminal justice, I like to present two examples of real-life-resembling situations. There is an inherent alignment with logical fallacies and the grounds on which the law enforcement penalties exist. tinyurl.com/logical-fall...

I would be uploading a youtube video in a few days, while I review my draft on First Order Logic. I am happy to provide any authorship verification of this document as needed. I also could not share many details during planning stage with anyone due to other demanding priorities.

With the advent of LLMs, spreading awareness on First Order Logic sems to have made it easier to self-learn using Human-LLM assisted learning. This draft tutorial is fully written and prepared by me. It was not reviewed by anyone. Please contact sakoju2014@gmail.com tinyurl.com/fol-as-a-tool

This is my initiative to spread awareness of First Order Logic as a tool for reasoning in news/social media for fact/logical verification. Logic has become an occult subject similar to many AI and computer science subjects, referring to public opinion.

- A made-up news premise, hypothesis with text and Image example from IBM and NewYork Times News provenance project. tinyurl.com/fol-as-a-tool
- Another made up example asks you "would you have believed the made up hypothesis as a news headline?"
tinyurl.com/fol-as-a-tool

Checkout real life examples (if you are already familiar with First Order Logic):
- some examplea served as an onigiri, but have logical fallacies similar to real life that often go undetected by humans. tinyurl.com/fol-as-a-tool

This tutorial also expects human learners to critique the responses of LLM/AI assiatant of First Order Logic forms for evaluation and practice. This module uses a lot of real world examples for using First Order Logic as a tool to detect logical gaps in reasoning from human-generated speech/texts.

In this module, I introduce a new learning approach where I provide training with real life examples and also I also ask learners to use an AI assistant as an expert to teach and evaluate First Order Logic to the human learners. tinyurl.com/fol-as-a-tool

similar to how research works used First Order Logic to evaluate logical fallacies and paradoxes etc in LLMs, Ontanon et all (2022), oshi, Saparov, et al and Lalwani et al.

I introduce First Order Logic (FOL) as a tool for real life complex reasoning situations such as logical fallacies, paradoxes, misinformation - - on social media or in your life. My draft tutorial to teach FOL using tinyurl.com/fol-as-a-tool .

With a Masters in Forensic Science and Criminal Justice, I've studied countless cases where impulsive words—online or in-person—create lasting damage for both the speaker and recipient. This guide applies those insights to everyday communication. tinyurl.com/transform-co... #CriminalJustice

While the ideas in my communication guide are my own tinyurl.com/transform-co..., the effective tweets promoting it were crafted with assistance from @AnthropicAI 's Claude. Great example of AI enhancing safe human communication on a public platform! #ClaudeAI

IMPORTANT: While AI assistants can help improve difficult conversations, they aren't dumping grounds for toxic language. Most platforms monitor for harmful content and may deactivate user accounts misusing these tools.

Learn ethical AI communication: tinyurl.com/transform-co... #AIEthics

Digital harassment often misrepresents who we really are. What appears online can't capture our full human experience—or what we deserve.

Learn how AI can help transform toxic interactions into meaningful dialogue: tinyurl.com/transform-co... #BetterConversations #AIforGood

Ever wished for the perfect words during a heated argument or uncomfortable confrontation? My new guide shows how AI assistants can transform toxic interactions into productive conversations for daily life. tinyurl.com/transform-co... #BetterConversations #AIforGood