Projects

Here at Sunrise Technology Inc. we are dedicated to the advancement of machine learning techniques for all applications. Some of our research/commercial projects are listed below.

Autonomous Driving:

Autonomous driving is the future of vehicles, we at Sunrise Technology Inc. are passionate about developing methodologies to improve the current models to realize this goal. Utilizing cutting edge technologies and leveraging the power of machine learning we aim to maximize the reliability and reduce the cost of autonomous vehicles for a wide variety of applications outside of just passenger transport.

Currently we are hosting a camp for Summer 2023 where students can come and learn machine learning and programming techniques from our in house engineers. Develop and deploy your own custom self-driving model! If you’re interested click below to learn more.

More Info
FAIR Data Management:

At Sunrise Technology, we are dedicated to revolutionizing the field of material science through the power of cutting-edge deep learning techniques. Our most recent project focuses on automating the analysis of x-ray scattering images, which provides crucial insights into the physical structure of materials at the molecular scale.

To achieve this, our team has developed innovative methods using Convolutional Neural Networks and Graph Neural Networks for x-ray scattering image classification. Recognizing the need for extensive training data, we employed simulation software to generate synthetic x-ray scattering images. By simulating various detectors, our model generates embeddings that are invariant to experimental setups, making it easier to search for similar materials imaged under different conditions.

Our experiments demonstrate that our deep learning-based approach is not only robust but also produces valuable cross-detector embeddings. Furthermore, our method attains high classification accuracy for essential material properties, solidifying its potential to transform the material science landscape.

At Sunrise Technology, we are committed to harnessing the potential of advanced technology to drive innovation and expand the frontiers of material science. We look forward to continuing our work and contributing to the development of new materials and technologies for a better future.

Laser Feedback Control:

Physicists working at large particle collider systems maintain a beam-line on which they run their experiments. To control the beamline, physicists rely on different types of classical control schemes. To make this adaptive and autonomous we developed a reinforcement learning agent to act as the stabilizer, controlling the laser and minimizing the noise. Our goal with this project is to deploy a robust and fast control system that can learn from a changing environment while also accounting for any high frequency noise.

Orbit Feedback Control:

National Synchrotron Light Source II is a third-generation storage ring producing synchrotron radiation through laser-electron interactions. Low emittance in a light source facility requires stable electron beam orbit. We aim to design an adaptive orbit feedback system based on modern data science methods and state-of-art deep reinforcement learning algorithms.

Particle Collision Triggering:

At Sunrise Technology, we are proud to contribute to groundbreaking research in the field of high-energy and nuclear physics. Our latest project, centered around the sPHENIX detector at Brookhaven National Laboratory, has led to the development of an innovative model architecture for trigger detection. This technology is crucial for enhancing data acquisition efficiency and streamlining downstream offline data analysis processes.

Our research focuses on detecting physics processes involving charm and beauty quarks, which provide valuable insights into the formation of the early universe. Utilizing geometric information from two fast silicon detectors and introducing transverse momentum as an intermediate feature, we’ve created a binary classification model that outperforms current state-of-the-art algorithms. By integrating a bipartite graph neural network with an attention mechanism, our model achieves a significant increase in accuracy and AUC score by more than 15%.

Sunrise Technology is dedicated to pushing the boundaries of scientific understanding through advanced technology and innovative solutions. We look forward to continuing our work in high-energy and nuclear physics experiments and contributing to a deeper understanding of the universe.