Dr. Kobi Abayomi’s Post

In the field of Artificial Intelligence (AI), the significance of data quality and access cannot be overstated, and Ikon Science leads this space with our cutting-edge software, Curate. Sushma Bhan, a non-executive board member, recently emphasized in an article that Curate is designed to consume, centralize, and apply AI to process thousands of documents, converting them into actionable insights. This process significantly improves data retrieval and decision-making within the energy industry. With Curate, we are revolutionizing the way users engage with enterprise data! Learn more: https://hubs.li/Q02sn_D50 #AIInnovation #IkonScience #Curate

Can deep learning networks approximate any continuous function? How many layers does one need? Well surprisingly a single hidden layer suffices to approximate any continuous function. We study the proof due to Cybenko (1989), which is specific to sigmoidal class of non linearities in the video below. We also explain the concepts from real analysis needed for the proof. #deeplearning, #ml, #universalapproxtheorem

Interesting read on Softmax that I stumbled on-- nothing directly actionable with my work presently, but I wouldn't be surprised if in the next <some near interval> we start seeing changes in Transformer models.    This resonates with me; my first introduction to the Softmax algo was in Statistical Thermodynamics (it's not called Softmax there, rather the Boltzmann distribution)--essentially the probability that a given particle will be in some thermodynamic state at a given system temperature. I remember being fascinated when the professors would say things like "this model represents a particle's most likely 'choice' of states to exist as a function of its systemic environment". But a non-choice is still a choice; this author proposes a way of encapsulating that. (Oh the analogies to life that can be made with that!) Attention is Off by One: https://lnkd.in/g3ZyU8-7

A Comprehensive Overview of Gaussian Splatting ... https://lnkd.in/eMQG7Kym #AI #ML #Automation

"Data science has revolutionized the oil and gas industry over the past 75 years, evolving from large computers in 1957 to the integration of AI and ML today. Industry leaders highlight the impact of technology advancements and the promising future of data science in the sector. Explore how this transformation can benefit your business. #DataScience #OilAndGas #AI #MachineLearning #IndustryEvolution" You can read more about it in the article: https://lnkd.in/dmzG3vBs

Physics-Informed with Power-Enhanced Residual Network: Results, Acknowledgments & References ... https://lnkd.in/eVC6BQnh #AI #ML #Automation

AI sci-fi story "Eclipse of Hope" chapter 1, see details on YouTube https://lnkd.in/gmt7mnXB #aigenerated #aivideo #aishorts  Text to video in seconds, using #aiTransformer #Speech_Synthesizer (https://lnkd.in/gVFGZxKe)

An entity's "sensitiveness" to a particular set of environmental patterns, that trigger the entity's corresponding action/behaviour, is the dynamo of the dialectic relationship that binds organisms to their environment and that is responsible for life itself. This principle is the grouding stone of biosemiotics and as I have been advocating for years should be the grounding stone of all Artificial Intelligent Systems theories if they are really interested in producing intelligent-like entities. Significant steps towards this approach seem to be present in A-SOID...

Todays recommended reading! I found this article discussing the use and development of cognitive technology in his risk sectors, such at the Norwegian petroleum industry. I'm also very excited and proud to see some of my former students pushing ahead in this field! Congrats, great work! The term 'cognitive technology' is interesting as it is more focused than the general term 'AI' on how tech can improve/shape typical cognitive tasks such as perceiving, learning, reasoning, planning and/or executing procedures. This might allow a more concrete discussion on how to use cognitive tech in high risk settings - and particularly opens up the questions of how we can use, develop and perhaps contain tech that exponentially changes itself. How do we use tech that we struggle to understand the nature of in setting with high stakes and catastrophic potential? How can we team up with AI and cognitive tech? #humanfactors #hfe #highriskAI #cognitivetechnology #humanAIteaming Jørgen Ernstsen Bjarte Knappen Røed, Ph.D Mica Endsley Gary Klein Kristian Gould Jan Tore Ludvigsen Georg Giskegjerde Maria Kinck Borén Mats Ekre Wang Koen van de Merwe Følstad Asbjørn Salvatore Massaiu Vegard Kolbjørnsrud https://lnkd.in/dCH_-6rm