»New paradigms for machine learning on non-volatile memory architectures«
Since many years, the Chair for Embedded Systems works internationally successful in the areas of computer engineering, such as smart embedded systems. Many interesting and open problems in these areas need to be addressed to successfully deploy such systems in modern application domains.
Embedded systems usually have limited energy, computing power, and memory/storage space. Hardware accelerators and revolutionary memory hierarchies are promising candidates to tackle these challenges. On the memory side, the emerging byte-addressable non-volatile memories (referred to as NVMs for short), such as Phase Change Memory, Spin-Transfer Torque RAM and Resistive RAM (ReRAM), feature low leakage power, high density, and low unit costs. NVMs are hence interesting alternatives to replace DRAM as main memory and/or hard disks and flash as storage. NVMs offer the opportunity for new memory architectures and open new computing paradigms like near-memory and memory computing.
In the scope of this research project, we want to investigate energy efficient machine learning applications on resource constraint devices/systems in which the underlying architectures profit from in-memory and/or near-memory computing.
Salary category 13 TV-L, depending on the fulfillment of professional and personal requirements.
limited for 1 year with the possibility for multi-year extension
January 1, 2025
For further information, please contact Prof. Henkel the mail topic: Application CES-ML.