Svilen Kanev
Near-Memory Processing (NMP) is an emerging technology that can help accelerate pointer-chasing applications. However, it is necessary to invoke further evaluation of NMP on workloads that are more complex than data structures. We identify garbage collection as a more complex application for NMP. In this paper, we demonstrate that we can get 2x performance improvement in short-lived processes and a 2.3x reduction in L2 traffic by integrating NMP into the built-in Java garbage collector. However, we will dis
Deep learning solutions become one of the most important intellectual properties of the solution providers. A few studies raised the alarm about the leakage of important model parameters through model extraction attacks. This paper discusses the practicality of such attacks by exploring the baseline assumptions of the existing attack scenarios. View Paper
Homomorphic encryption (HE) is a privacy-preserving technique that enables computation directly over ciphertext. Unfortunately, a key challenge for HE is that implementations can be impractically slow and have limits on computation that can be efficiently implemented. For instance, in Boolean constructions of HE like TFHE, arithmetic operations need to be decomposed into constituent elementary logic gates to implement so performance depends on logical circuit depth. For even heavily quantized fixed-point ar
Svilen Kanev