Introduction
This document serves as a formal erratum for the paper "CASH-RAM: Enabling In-Memory Computations for Edge Inference Using Charge Accumulation and Sharing in Standard 8T-SRAM Arrays," originally published in the IEEE Journal on Emerging and Selected Topics in Circuits and Systems in September 2020. The purpose of this correction is to ensure accurate academic attribution for the lead researchers involved in the study.
The Correction
The primary correction concerns the recognition of equal contribution among the lead authors. In the original publication, the specific roles and the shared nature of the primary research efforts were not explicitly detailed in the author metadata.
The erratum formally establishes that Amogh Agrawal and Adarsh Kosta contributed equally to this work. This designation of co-first authorship reflects the collaborative effort required to develop the CASH-RAM architecture, which involves complex cross-disciplinary work in hardware design and algorithmic optimization. Additionally, Amogh Agrawal is designated as the corresponding author for the research conducted at Purdue University.
Context of the Original Research
While this document is administrative in nature, it pertains to a significant advancement in edge computing. The original CASH-RAM paper addressed the "memory wall" challenge—the energy inefficiency caused by moving data between memory and processors in AI applications. The proposed CASH-RAM (Charge Accumulation and Sharing RAM) utilizes standard 8T-SRAM arrays to perform In-Memory Computing (IMC), significantly reducing the energy footprint for AI inference on edge devices.
Conclusion
Accurate attribution is vital for the integrity of the scientific record and the career progression of researchers. This erratum ensures that both Amogh Agrawal and Adarsh Kosta receive proper credit for their foundational roles in the CASH-RAM project. The scientific record now accurately reflects the collaborative nature of this hardware-software co-design effort.
Summary updated