CAIR Centre for Artificial Intelligence Research

#OpenPositions join the #TsetlinMachine team in developing next-generation #cybersecurity solutions for #IoT! Postdoc: https://lnkd.in/db79fTmF Integrated PhD: https://lnkd.in/dzdey6xh Deadline: December 4, 2024. The positions are connected to the NRC project "SecureIoTM: Energy-efficient intrusion detection systems for IoT based on logic-based Setlin machines". The goal of SecureIoTM is to combine Setlin machines, intrusion detection systems, and the Internet of Things (IoT) to develop the ultra-energy-efficient detection mechanism to combat rising IoT-related cybercrime. The project will develop innovative Tsetlin machine mechanisms and hardware designs to enable real-time anomaly detection and learning on resource-constrained IoT devices. https://lnkd.in/dYGfJyef #AI #ML #MachineLearning #Security #GreenAI #LogicalAI #DemocraticAI

Congratulations!!!

In limit orderbook data, we empirically show the larger the spread, the larger the microprice adjustment, which is consistent with the profitability of market making.  Furthermore, the further the orderbook is skewed towards higher supply (ask side) the more positive microprice adjustments occur. In contrary, the higher the true demand (bid side), the more negative adjustments are made. However, extreme imbalance on either side implies less adjustments to the microprice, most likely due to adverse selection.  Java code for building microprice estimates from orderbook data found here https://lnkd.in/dCEBYnZs

My recent paper "High resolution microprice estimates from limit orderbook data using hyperdimensional vector Tsetlin Machines" out now. Here we show TsetlinMarketViewFX demonstrating on TSLA the development of the "cone" shape in microprice prediction in realtime, implying that larger the spread (x-axis), the larger the microprice adjustment (y-axis), which is consistent with the profitability of market making. The second part of the video shows how microprice adjustment predictions are correlated with the orderbook imbalance at N price rank levels (here we use N=5). Namely, the more the orderbook is skewed towards higher supply (ask side) the more the positive microprice adjustments occur. In contrary, the higher the true demand (bid side), more negative adjustments are made. However, extreme imbalance on either side implies less adjustments to the microprice, most likely due to adverse selection. 

CAIR's Bimal Bhattarai is defending his PhD thesis next Monday. I am deeply impressed by his research and achievements in advancing the Tsetlin machine research field. Best of luck, Bimal! Thesis: https://lnkd.in/duD-GsW2 Disputation: https://lnkd.in/d78khxDM

Tonight marks the International Researcher Grand Prix, featuring the exceptional PhD candidate Henrik Brådland. I encourage everyone to tune in to the live stream from 19:00 to 21:00 at https://lnkd.in/dbg_n9HX . Henrik will be presenting on the impact of AI in transforming the public sector and streamlining building permit processes. You can participate by voting and ranking the presenters via Slido, using event code 3118 602 between 19 and 21 tonight. Let's show our support for Norway, UiA and Henrik—I'm confident that after watching his talk, he will stand out as one of the top communicators of the evening! Every vote brings Henrik closer to victory, so please feel free to share.

Spennende artikkel om våre venner i Literal Labs som bygger fremtidens KI ved hjelp av Tsetlinmaskinen! #bærekraftig #forklarbar #grønn #rettferdig #KI #ByeByeBlackBoxes https://lnkd.in/dBEXb2Ti Noel Hurley Xavier Parkhouse-Parker Rishad Shafik Alex Yakovlev

#GreenAI #DemocraticAI #ScientificAI #ByeByeBlackBoxes

Supercomputing at CAIR At the Centre for AI Research (CAIR), we are proud to leverage cutting-edge NVIDIA DGX systems that excel in both AI training and inference. Our advanced computing resources are designed for flexibility, allowing multiple projects to run concurrently, which positions CAIR as a premier AI research hub in the Nordics. Located at UiA Campus Grimstad, our supercomputing capabilities include: 208 petaFLOPS of total computing power. 8 NVIDIA DGX-2 systems configured in a high-speed 100Gbit InfiniBand cluster, delivering 16 petaFLOPS for typical AI tasks (FP8). 6 NVIDIA DGX H100 systems offering an astounding 192 petaFLOPS for demanding AI applications (FP8). With this unparalleled computational power, CAIR is not only advancing research but also empowering innovative solutions that will shape the future of AI. Join us as we lead the way in transformative AI research!

#Announcing the first #deep #tsetlinmachine! Reason #logically. Reason #multimodally (images, text, measurements, etc.). Reason #understandably. Reason #efficiently. Reason #safely. https://lnkd.in/dY7QgmkX #StayInCharge #GreenAI #DemocraticAI #ScientificAI #ByeByeBlackBoxes