Case Studies
Our explainable AI assessments provide you with AI solutions for critical operations that you can trust.
Efficient Monitoring of Physical Devices in Infrastructure
Large physical infrastructure sites often have numerous devices with meters or dials that require monitoring. Collecting these readings can be expensive, time-consuming, or overlooked altogether. Zetane provides an efficient solution for detecting and reading physical meter devices, needed to maintain remote sites, without costly physical visits and provides real-time monitoring and alerts.
Learn more
Ensure Safe and Reliable AI Models with Comprehensive Testing and Validation
Predicting AI model behavior in out-of-domain scenarios can be challenging, often leading to unexpected failures in unknown situations. Thorough testing and validation of AI models are essential for applications with real-world consequences, but this process differs significantly from traditional software testing. Zetane's Protector Platform enables testing numerous scenarios simultaneously, allowing users to observe model behavior and understand boundary conditions. By augmenting models with edge case data, the platform significantly improves model robustness and overall performance.
Learn more
Analysing Aircraft Black Box Data for Predicting Adverse Effects
A world-leading company specializing in the analysis of aircraft flight data recorders (FDRs) wanted to leverage the power of machine learning to analyze the massive amounts of data from FDRs and identify safety-critical events. Zetane's team of industrial AI experts worked on the project from data preparation to model design and training to debugging and optimization. The client used the resulting project to apply for an important R&D grant.
Object Detection and Model Testing for Autonomous Trains
A global aerospace and transportation client faced a tight deadline to demonstrate the results of an object detection project. With the critical deadline looming, Zetane worked with the client to understand and explain how their data was influenced the machine learning model. This identified a major flaw in the model that was rectified on time and provided a remarkable and successful demonstration to decision-makers.
Labelling and Analysis of Satellite Images
Presagis, a world leader in commercial-off-the-shelf-software (COTS) for modeling and simulation developed a working prototype of a machine learning solution to label roof-tops from satellite images. Decision makers initially did not support the project and it received little interest. An important conference for the company was weeks away and using excel spreadsheets, graphs and powerpoints would be insufficient. Zetane worked with the team to display the project using the Zetane Engine, where the rich visuals brought their work to life. Now convinced, management decided to use the Zetane visual demo at the conference, which received accolades from conference attendees.
Have more questions?
Contact Us
Predictive Maintenance For Heavy-Duty Vehicles
A global defense company that manufactures heavy transport vehicles wanted to demonstrate the use of machine learning to predict the future maintenance needs and the failure of critical systems for these vehicles. They made good progress but struggled to bring the project to a working MVP and had a difficult time explaining how the AI-predictive-maintenance system would function in an operational context. Zetane industrial AI experts assisted the clients' AI team to: understand and explain the inner workings of the model; pinpoint where improvements could be made; created real-time dashboards that operators could understand; and created a demo that impressed the Board of Directors. No doubt, the successful MVP received further funding to scale the project.
Human-Understandable Demo of ML Solution for Energy Grid Optimization
An up-and-coming company that uses machine learning to optimize the use of renewable sources of energy by cities, was struggling to convey the benefits of their groundbreaking work to potential clients. With an important global conference on renewable energy fast approaching, the company tasked Zetane to convert their complex data and AI technology into a powerful and convincing demonstration that showed cost-savings and environmental benefits of their solution. This intuitive demo helped secure new municipal customers.
Predicting Patient Outcomes of Brain Surgery Using Simulators
Canada's National Research Council developed a simulator to train surgeons in complex brain surgery procedures and had amassed large quantities of training data in the process. Their research team developed machine learning models with the data to predict patient outcomes from the movements of surgeons in the simulator. They needed to validate their results and demonstrate their findings. Zetane stepped in to bring their research to life as detailed, intuitive visuals that generated 3D simulations of the operations linked to the ML models. In addition to greater transparency of their findings provided by our powerful demo, the visuals by Zetane also helped the researchers identify and fix notable errors in their model.
Zetane Videos
Case studies videos where we helped customers bring their AI solution to the next phase.
Autonomous Train Proof of Concept
Satellite Imagery Explainable AI
Flight Data Analysis AI Dashboard
Zetane enables us to see inside the black box of the neural networks we design and to see the effect of each of the inputs on the outputs, which improves our ability to explain the results.
Rola Harmouche, PhD, Research Officer in Machine Learning, National Research Council of Canada