Lifelong learning—an agent's ability to learn throughout its lifetime—is a hallmark of
biological learning systems and a central challenge for artificial intelligence (AI). The …

The posit representation for real numbers is an alternative to the ubiquitous IEEE 754
floating-point standard. In this work, we present PERCIVAL, an application-level posit RISC …

To accelerate the inference of deep neural networks (DNNs), quantization with low-bitwidth
numbers is actively researched. A prominent challenge is to quantize the DNN models into …

Deep neural networks (DNNs) have become an enabling component for a myriad of artificial
intelligence applications. DNNs have shown sometimes superior performance, even …

Spiking neural network models when deployed in dynamic environments, catastrophically
forget previously learned tasks. In this paper, we propose a reconfigurable spiking digital …

While vision transformers (ViTs) have shown great potential in computer vision tasks, their
intense computation and memory requirements pose challenges for practical applications …

The posit representation for real numbers, aka Unum-v3, is an alternative to substitute the
IEEE 754 standard and thus mitigate the inherent problems to the construction of floating …

In this paper, a new low-precision hardware-software codesign framework is presented, to
optimally select the numerical formats and bit-precision for TinyML models and benchmarks …

Neural network models catastrophically forget previously learned information while
acquiring new knowledge, requiring a fundamental change in learning models and …

For many decades, IEEE floating-point formats are used as the golden numeric formats for
many applications including signal processing, linear algebra evaluation, and other …