Aleksandar Zlateski

Title: Innovator Profile: Aleksandar Zlateski

Introduction: Aleksandar Zlateski is a notable inventor based in Somerville, MA, recognized for his contributions to the field of neural networks. With a total of two patents to his name, Zlateski is making significant strides in the realm of artificial intelligence and machine learning.

Latest Patents: Zlateski's latest patents focus on "Systems and methods for generation of sparse code for convolutional neural networks." This innovative approach provides a system and method for generating code aimed at executing neural networks, particularly convolutional neural networks (CNNs) which might contain multiple convolutional layers. The patent details a process where, for each non-zero element in a kernel tensor or matrix linked to a convolutional layer, specific instructions are generated. This includes creating vector broadcast instructions and fused multiply-add (FMA) instructions that utilize registers to manage input data and store outputs necessary for convolution operations. The software developed can be executed during convolutional tasks, potentially impacting various applications, including NN inference applications.

Career Highlights: Aleksandar Zlateski has established himself as an innovator in the AI sector, currently employed at NeuralMagic Inc. His work focuses on enhancing the efficiency of convolutional operations within neural networks, contributing to advancements in AI technologies.

Collaborations: During his career, Zlateski has collaborated with his coworker, Justin Kopinsky. Their joint efforts have contributed to the development of cutting-edge technologies in the field, helping to push the boundaries of what is possible with neural networks.

Conclusion: Aleksandar Zlateski's innovative contributions and his role at NeuralMagic Inc. underscore the importance of advancing techniques in artificial intelligence. With his two patents focusing on neural network efficiencies, Zlateski continues to influence the technological landscape, paving the way for future innovations in machine learning and computational efficiency.