Georgina Cosma
@gcosma1
Followers
16K
Following
2K
Media
856
Statuses
3K
Professor of AI @lborouniversity Neural Information Processing, Retrieval & Modelling, NLP, LLM Learning & Unlearning, AI Ethics, AI in Health &Social Care
England, United Kingdom
Joined December 2010
🚀 Excited to share my new academic website: https://t.co/VLOjVWOCv4 ! It features my research, key projects, publications, and ongoing work at the intersection of AI and science. Take a look! #AI #DataScience #ProjectShowcase
1
0
8
Massive congratulations to my PhD student Mikel W-Lekuona for passing his viva with flying colours! 🎓✨ "Cross-Modal Image-Text Retrieval for Real-World Applications: Advancing Learning, Evaluation, and Benchmarking Methodologies" #AI #InformationRetrieval
1
0
4
Two fantastic events in one week! 📍Delivered “Preserving and Accessing: Secure AI Solutions for Sensitive Information Systems” in Manchester on RAG-based approaches. 💡Spoke on LLMs &Generative AI at the Pack to the Future conf@society_east, hosted at #SDCA @lborouniversity.
0
0
2
Just tested a serious #AI vulnerability: hidden white-text prompts in a doc were executed without detection by both ChatGPT(@OpenAI) and Claude(@AnthropicAI). No resistance, no safeguards. #LLMs remain highly susceptible to doc-based prompt injection. This needs urgent attention.
1
3
7
"Tell me and I forget, teach me and I may remember, involve me and I learn"" Benjamin Franklin
27
135
848
Nice tutorial! #Agentic #RAG: Step-by-Step Tutorial W/ Demo Project. Learn how to build an Agentic RAG pipeline from scratch, integrating local data sources and web scraping to generate context-aware responses to user queries. #LLMs #NLP
https://t.co/p6k5qV8OVe
0
0
3
An exciting & engaging few days at #HealTAC2025 in Glasgow! Great to see academic, clinical, industry, & patient communities come together to advance healthcare analytics. Thought-provoking discussions on #freetext & #multimodal data. Already looking forward to #HealTAC2026! #NLP
#HealTAC2025 starting in Glasgow, with an introduction by Peter Harrison from @HDR_UK #HealthTech #LLMs #datasaveslives
0
0
2
🎓Insightful day at #healtac2025 on #NLP in #healthcare!🚀Presented our #AI-powered multi-framework concept annotation tool for Prevention of Future Death reports-automated safety intelligence extraction w/dashboards, augmenting existing incident reporting https://t.co/2fjbaht1dx
0
1
2
Nicely written piece by James Coe on REF2029 guidance 📚 "REF is about institutions not individuals" - key insight for understanding the upcoming framework 🎯 #REF2029 #HigherEd
https://t.co/N0wPvmL3hH
0
1
2
📢Published! My PhD student working on cross-modal learning has published: "FiCo-ITR: bridging fine-grained and coarse-grained image-text retrieval for comparative performance analysis" 📝OA paper: https://t.co/sZXlv922dn 💻GitHub (Py code+tutorial): https://t.co/1VoAIReQRG
#AI
0
0
7
📢Thinking about an MSc in AI or Data Science? 🌟 Boost your career with cutting-edge, practical skills that top employers want!👉 Check out our courses & world-class sports campus!🚀MSc AI: https://t.co/GQPWYBssVu 📊MSc Data Science: https://t.co/S1ewB65WKp
#AI #DataScience
0
1
6
🎓Fully funded Sanctuary PhD Scholarships for at-risk scholars. I'm looking for #PhD students in #NLP (unlearning for LLMs & other topics) - get in touch if you're eligible &interested! 🗓️Deadline for Oct2025/Jan2026 entry: 23 June2025 🔗 https://t.co/ILZxeyq13n
#phdscolarship
0
2
7
Weekend fun!🎉 #Gemma3, Google's cutting-edge multimodal models(text+ image) come in 1B, 4B, 12B, and 27B sizes.🚀Plus, finetune #Llama 3.3, Gemma 3, Phi-4, Qwen 2.5 & Mistral 2x faster with 80% less VRAM! 💡 Github w/ ipynbs https://t.co/gXO1RHVslD Colab
github.com
Fine-tuning & Reinforcement Learning for LLMs. 🦥 Train OpenAI gpt-oss, DeepSeek-R1, Qwen3, Gemma 3, TTS 2x faster with 70% less VRAM. - unslothai/unsloth
0
0
13
AI can write code, but it can't understand consequences. 🤖 Human programmers provide judgment, ethics, and accountability for systems that affect real lives. 🧠 Humans work alongside AI, not get replaced by it. 🤝 #HumansNeeded 🔍 #AI
Some people today are discouraging others from learning programming on the grounds AI will automate it. This advice will be seen as some of the worst career advice ever given. I disagree with the Turing Award and Nobel prize winner who wrote, “It is far more likely that the
0
0
1
🚀 Recently read this v. important paper: "Deduplicating Training Data Makes Language Models Better." & I delivered a lecture on this!📚✨Deduplication is a must for better NLP models! Paper: https://t.co/583nMBlN8p My slides on the paper: https://t.co/O9ratLR8N0
#LLMs #AI #NLP
0
1
4
📢Thinking about an MSc in AI or Data Science? 🌟 Boost your career with cutting-edge, practical skills that top employers want!👉 Check out our courses & world-class sports campus!🚀MSc AI: https://t.co/GQPWYBssVu 📊MSc Data Science: https://t.co/S1ewB65WKp
#AI #DataScience
0
1
6
Another year, another great MSc #AI cohort! 🎓 In my #NLP module, we covered vector space models & neural approaches 🤖, info retrieval 🔍, topic modelling 📊, summarisation ✂️, deduplication 🔄& more. Students tackled real-world challenges! 🚀 MSc AI:
https://t.co/GQPWYBssVu"
0
0
1
🚨In the media! Our AI model can predict how long a person with a learning disability & MLTCs is likely to stay in hospital! Some articles: [LU] https://t.co/5oudlDDq25
#healthcareAI #machinelearning [LTD] https://t.co/oZ4ZuqeGK8
@AIM_RSF @NIHRresearch @gcThomasJun @Dr_S_Kumar_G
0
0
6
Published! Equitable hospital length of stay prediction for patients with learning disabilities and multiple long-term conditions using machine learning #healthcareAI #machinelearning @DECODEproject2 @AIM_RSF @NIHRresearch @gcThomasJun @Dr_S_Kumar_G Paper: https://t.co/E3HqQZPhsd
1
3
9
🚀 Pushing the boundaries of AI with these self-funded PhD projects in🔬 Unlearning and LLMs: 🔬Machine Unlearning https://t.co/1cSuoYnaR1
https://t.co/uyPIs1B8RG 🔄Continual Adaptive Learning https://t.co/rtZuJZn5hL 📊Semantic-Aware Data Deduplication
0
1
2