«I'm sorry Dave. I'm afraid I CAN do that.»
Beyond the hype, projections of utopian worlds or dystopian fears inspired by fictional narratives, Artificial Intelligence is in reality a simple tool, a class of algorithms among others, allowing to bring practical and real answers for complex issues in a wide variety of areas.
Where a non-trivial problem has to be optimized, where a statistical answer has to be found in a complex dataset, where a process can be automated, Artificial Intelligence can add value by finding innovative solutions or providing a better adequacy to the problem.
Our mission at Tentacle Devs, is to analyse and find the best match between the context and the state-of-art over 50 years of Artificial Intelligence research to create technology with real and operational results.
Founded in 2016, Tentacle Devs is however not a startup riding the wave of hype, and cumulates 20 years of experience on myriad variations of AI systems. Having a wide range approach, Tentacle Devs builds its own technology but also provides service for consulting, analysis, specific development, study case, data science, POCs, prototypes.
There are many subclasses of Artificial Intelligence and there is not just Deep Learning to adequately respond to a problem. Tentacle Devs has the expertise to answer and set up complex systems, by analyzing and addressing as accurately as possible the needs, without excess and without underestimating the dimension of the problem.
Analyzing each case study carefully as a unique case, we make sure to set realistic expectations and deliver on promises that are not based on the promise-all-hype. For instance, in healthcare, we can work on image-based assistive diagnosis tools but not on best treatment prescription tools for cancer patients.
Many AI systems learning from data, are designed and developed with little knowledge of mathematical or human biases that can be introduced into AI design, datasets, programming, or parameterization. The developed system then goes seemingly through the steps of validating, while it has errors within it, that make it potentially unsuitable, falsely generalizing, or dangerous in a real environment. We have the practical experience and the methods to fight against various forms of hidden biases that are not described in the litterature, and we implement them for the design of developed technologies.
There is a large panel of fields able to take benefits of what AI can bring in the operational ground. In each category, here are a few examples of technologies we can handle.
Artificial art, AI painting, historical film colorization, style transfer, stable diffusion
Realistic simulation or opponents, automative lipsynching from audio, human-like behavior imitation, dynamic tree-dialog-based language for adventure games, in-game evolving characters, automative game builder/balancing with player feedback
Evolutionnary locomotion/behaviors, automated reasoning, robot path planning in warehouses, Artificial Life, bio-inspired cooperative scheduling, sport robots
Soft driving, multicars path planning, Reinforcement Learning, End-to-end, scheduling autonomous traffic from traffic lights
Automated image analysis, denoising and enhancing medical image data, autonomous assistive diagnostic tools.
Pixel-based detecting statistical anomalies, physics-based and geometry-based anomalies detection tools
Internal abnormal activity, external intrusive attacks like bruteforce, spydering, injections...
Potential conflicts or threats in crowd videos
Optimization of green energy source extraction controls. Distributed energy spots scheduling
Optimization of stocking and destocking process for automated warehouses.
-C/C++/C# -Python -AI frameworks (TensorFlow, Caffe, dlib, Pytorch...) -AI clouds (AWS, Azure cognitive services...) -Neural networks -DNN -RNN -CNN -Reinforcement Learning -Qlearning -Genetic Algorithms -Cellular Spatial GA -GAN -Autoencoders -Planification -State machines -Dialog Trees -Multithreading -Artificial Art -Multiscaling -Computer game bots -Game theory and strategy -Image recognition -Face recognition -Speech recognition -Automated reasoning -Artificial Life -Bio-inspired computing -Crowd simulation -Knowledge representation -AI compilers -Realtime Pathfinding
