BigData_UGent (@bigdata_ugent) 's Twitter Profile
BigData_UGent

@bigdata_ugent

Big Data Hadoop, HDFS, Spark, ML Flow initiative at Ghent University

ID: 3245178101

linkhttp://www.bigdata.UGent.be calendar_today10-05-2015 18:20:27

250 Tweet

123 Followers

156 Following

Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Great talk by Sander Dieleman (Sander Dieleman, Google Deepmind and @UGent alumnus) at #NeurIPS2024 Workshops in Vancouver, BC (Canada) on autoregression vs diffusion for iterative refinement. Visit his great blog: sander.ai #NeurIPS #AI

Great talk by Sander Dieleman (<a href="/sedielem/">Sander Dieleman</a>, Google Deepmind and @UGent alumnus) at #NeurIPS2024 Workshops in Vancouver, BC (Canada) on autoregression vs diffusion for iterative refinement. Visit his great blog: sander.ai  #NeurIPS #AI
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Incorporating context into forecasting improves performance on average. Llama 3.1 405b seems to work best. Great talk by Valentina Zantedeschi (ServiceNow) at the time-series workshop at #neurips24 #NeurIPS #NeurIPS2024 #AI #LLM

Incorporating context into forecasting improves performance on average. Llama 3.1 405b seems to work best. Great talk by Valentina Zantedeschi (ServiceNow) at the time-series workshop at #neurips24 #NeurIPS #NeurIPS2024 #AI #LLM
CVAMO (@cvamo_ugent) 's Twitter Profile Photo

Today, we are at the 2025 Flanders Make Scientific Conference in Leuven. CVAMO also has several posters at the event. It’s a great event for scientific networking. #FlandersMake #ORMS #MANPROCAD #REMASC #CVAMO

Today, we are at the 2025 Flanders Make Scientific Conference in Leuven. CVAMO also has several posters at the event. It’s a great event for scientific networking. #FlandersMake #ORMS #MANPROCAD #REMASC #CVAMO
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Great talk by Matthew A. Lanham on the #data4good competition at INFORMS’ #MAPD 2025 (Meeting of Analytics Program Directors at 2025Analytics) in Indianapolis, IN #predictiveanalytics #analytics #machinelearning #INFORMS

Great talk by <a href="/MatthewALanham/">Matthew A. Lanham</a> on the #data4good competition at <a href="/INFORMS/">INFORMS</a>’  #MAPD 2025 (Meeting of Analytics Program Directors at <a href="/2025_Analytics/">2025Analytics</a>) in Indianapolis, IN #predictiveanalytics #analytics #machinelearning #INFORMS
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Great application of optimization techniques in sports: USA Cycling (USA Cycling) has been awarded the 2025 INFORMS Edelman Award at 2025Analytics in Indianapolis, IN, the “Nobel Prize” of Operations Research, for its innovative use of O.R. and analytics in sports performance.

Great application of optimization techniques in sports: USA Cycling (<a href="/usacycling/">USA Cycling</a>) has been awarded the 2025 <a href="/INFORMS/">INFORMS</a> Edelman Award at <a href="/2025_Analytics/">2025Analytics</a> in Indianapolis, IN, the “Nobel Prize” of Operations Research, for its innovative use of O.R. and analytics in sports performance.
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Great panel discussion during the UGent FEB Research Day 2025 with representatives from Flanders Make, KPMG, OMP, UGent Industrial Research Fund. #PhDResearch #IndustrialResearchFund #Startups #GhentUniversity CVAMO #ecosystem UGentFEB

Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Last week, the 2025 Edition of our “Current Data Science for Business Students Meet Alumni” Event took place at UGentFEB (Ghent University). #ORMS #DataScience #DataAnalytics linkedin.com/pulse/current-…

Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

We have lift-off at the 2025 INFORMS Marketing Science Conference in Washington, DC. Hot topics include #GenAI, #LLM, #RAG (Retrieval Augmented Generation), #Marketing #Analytics, #ecommerce, #ORMS

We have lift-off at the 2025 <a href="/INFORMS/">INFORMS</a> Marketing Science Conference in Washington, DC. Hot topics include #GenAI, #LLM, #RAG (Retrieval Augmented Generation), #Marketing #Analytics, #ecommerce, #ORMS
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Today marks the start of presentations of the “Data Science for Business” (@UGent) projects. This particular project is for #FlandersMake using #RAG (Retrieval-Augmented Generation) #LLM technology to build a #chatbot. mma.ugent.be #orms #DataScience

Today marks the start of presentations of the “Data Science for Business” (@UGent) projects. This particular project is for #FlandersMake using #RAG (Retrieval-Augmented Generation) #LLM technology to build a #chatbot. mma.ugent.be #orms #DataScience
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

We have liftoff at the 2025 KDD Conference in Toronto, ON. #KDD2025 I’m attending the Workshop on AI for Supply Chain relevant to our REMASC. Supply chain planning faces two sources of complexity: radical uncertainty (e.g., use scenario planning) and need for good predictions.

We have liftoff at the 2025 KDD Conference in Toronto, ON. #KDD2025  I’m attending the Workshop on AI for Supply Chain relevant to our REMASC. Supply chain planning faces two sources of complexity: radical uncertainty (e.g., use scenario planning) and need for good predictions.
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Building scenario planning solutions for supply chain optimization using LLMs (large language models) at the #KDD2025 Workshop AI for Supply Chain. #FlandersMake #REMASC #AI See example on the slides in the pictures.

Building scenario planning solutions for supply chain optimization using LLMs (large language models) at the #KDD2025 Workshop AI for Supply Chain. #FlandersMake #REMASC #AI See example on the slides in the pictures.
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Great tutorial on efficient and effective search-agent training at #KDD2025 in Toronto. One does not need a lot of data to train a search agent via RL (using S3 compared to R1). #LLM #RAG #RL #R1 #S3

Great tutorial on efficient and effective search-agent training at #KDD2025 in Toronto. One does not need a lot of data to train a search agent via RL (using S3 compared to R1). #LLM #RAG #RL #R1 #S3
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

My personal #KDD2025 Day 2 is all about #finance ranging from using AI to enhance safety, AI in credit, and data-driven pricing #AI #llm #timeseries #finetuning #dataleakage

My personal #KDD2025 Day 2 is all about #finance ranging from using AI to enhance safety, AI in credit, and data-driven pricing #AI #llm #timeseries #finetuning #dataleakage
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Today, I started with the Test of Time Award talk at the #KDD2025 Conference in Toronto. Superb talk by Hao Wang on Deep Recommender Systems (RECSYS). He gave a great overview of how the field evolved over time, and the next steps. #AI #RecSys

Today, I started with the Test of Time Award talk at the #KDD2025 Conference in Toronto. Superb talk by Hao Wang on Deep Recommender Systems (RECSYS). He gave a great overview of how the field evolved over time, and the next steps. #AI #RecSys
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Great talk by Dan Roth on the limitations of using LLMs for reasoning at the #KDD2025 Conference in Toronto. Being linguistically coherent is much easier than we thought (that’s what LLMs have taught us). Use LLMs to generate (Python) code to solve hard problems. #AI #LLMs #ORMS

Great talk by Dan Roth on the limitations of using LLMs for reasoning at the #KDD2025 Conference in Toronto. Being linguistically coherent is much easier than we thought (that’s what LLMs have taught us). Use LLMs to generate (Python) code to solve hard problems. #AI #LLMs #ORMS
Dirk Van den Poel (@dirkvandenpoel) 's Twitter Profile Photo

Great talk by Sumit Gulwani (Sumit Gulwani, Microsoft) on improving AI reasoning by using intent, interaction, inspection, and memory (see last picture) at #KDD2025 in Toronto, Canada. #AI #LLM #CodeGeneration

Great talk by Sumit Gulwani (<a href="/SumitGulwani/">Sumit Gulwani</a>, Microsoft) on improving AI reasoning by using intent, interaction, inspection, and memory (see last picture) at #KDD2025 in Toronto, Canada. #AI #LLM #CodeGeneration