Raghav Singhal
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Raghav Singhal has a total of 13 co-authors. Here are the top co-authors by number of shared publications:
- Laxmeesha Somappa
IIT Bombay
2 shared publications - Maryam Shojaei Baghini
Department of Electrical Engineering, IIT Bombay, Mumbai, India
2 shared publications - Shreyas Deshmukh
Department of Electrical Engineering, Indian Institute of Technology Bombay, Mumbai, India
2 shared publications - Udayan Ganguly
Indian Institute of Technology Bombay
2 shared publications - Vivek Saraswat
Department of Electrical Engineering, Indian Institute of Technology Bombay, Mumbai, India
2 shared publications - Abhishek Kadam
Department of Electrical Engineering, Indian Institute of Technology at Bombay, Mumbai, India
1 shared publication
Latest Publications
- Open Access
- 9 October 2025
- Published by Association for Computing Machinery (ACM) in ACM Journal on Emerging Technologies in Computing Systems
- Vol. 21 (4) , 1-18
Analog and Temporary On-chip Memory for ANN Training and Inference
On-chip training at the edge becomes a primary requisite for real-time and security-sensitive artificial neural network (ANN) applications. In-memory computation (IMC) techniques have been proposed to facilitate data-intensive computational operations in ANNs. IMC-based multiply-accumulate (MAC) accelerates ANN training but suffers from significant communication overhead between the MAC engine and the off-chip storage for the intermediate data. This article proposes an analog temporary on-chip memory (ATOM) to store this intermediate data during ANN training. The ANN training architecture with the proposed ATOM has two significant advantages. First, the energy required to store intermediate data is scaled down by \(\sim\) 40 \(\times\) due to the on-chip and analog nature of the memory. Second, the proposed architecture avoids power and area-consuming analog-to-digital converters (ADCs) between neural network stages. The ATOM cell measurements are carried out from 20 fabricated chips, and the impact of ATOM characteristics on ANN system performance accuracy is analyzed. This article shows significant latency improvement of \(\sim\) 9 \(\times\) and area savings of \(\sim\) 5 \(\times\) for intermediate data storage compared to the on-chip SRAM during ANN training’s forward and backward pass operations. An improvement in the area and latency will be beneficial to instrument the area- and energy-efficient hardware system for on-chip ANN applications.- 9 October 2025
- Published by Association for Computing Machinery (ACM) in ACM Journal on Emerging Technologies in Computing Systems
- Vol. 21 (4) , 1-18
Open AccessAnalog and Temporary On-chip Memory for ANN Training and Inference
On-chip training at the edge becomes a primary requisite for real-time and security-sensitive artificial neural network (ANN) applications. In-memory computation (IMC) techniques have been proposed to facilitate data-intensive computational operations in ANNs. IMC-based multiply-accumulate (MAC) accelerates ANN training but suffers from significant communication overhead between the MAC engine and the off-chip storage for the intermediate data. This article proposes an analog temporary on-chip memory (ATOM) to store this intermediate data during ANN training. The ANN training architecture with the proposed ATOM has two significant advantages. First, the energy required to store intermediate data is scaled down by \(\sim\) 40 \(\times\) due to the on-chip and analog nature of the memory. Second, the proposed architecture avoids power and area-consuming analog-to-digital converters (ADCs) between neural network stages. The ATOM cell measurements are carried out from 20 fabricated chips, and the impact of ATOM characteristics on ANN system performance accuracy is analyzed. This article shows significant latency improvement of \(\sim\) 9 \(\times\) and area savings of \(\sim\) 5 \(\times\) for intermediate data storage compared to the on-chip SRAM during ANN training’s forward and backward pass operations. An improvement in the area and latency will be beneficial to instrument the area- and energy-efficient hardware system for on-chip ANN applications. - 5 July 2023
- Published by Elsevier in Neural Networks
- Vol. 165, 1050-1057
Enhanced regularization for on-chip training using analog and temporary memory weights
- 5 July 2023
- Published by Elsevier in Neural Networks
- Vol. 165, 1050-1057
Enhanced regularization for on-chip training using analog and temporary memory weights
- Open Access
- 11 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 79-85
Translation and Scale Invariance for Event-Based Object tracking
Without temporal averaging, such as rate codes, it remains challenging to train spiking neural networks for temporal regression tasks. In this work, we present a novel method to accurately predict spatial coordinates from event data with a fully spiking convolutional neural network (SCNN) without temporal averaging. Our method performs on-par with artificial neural networks (ANN) of similar complexity. Additionally, we demonstrate faster convergence in half the time using translation- and scale-invariant receptive fields. To permit comparison with conventional frame-based ANNs, we base our results on a simulated event-based dataset with an unrealistic high density. Therefore, we hypothesize that our method significantly outperform ANNs in settings with lower event density, as seen in real-life event-based data. Our model is fully spiking and can be ported directly to neuromorphic hardware.
- 11 April 2023
- Published by Association for Computing Machinery (ACM)
- p. 79-85
Open AccessTranslation and Scale Invariance for Event-Based Object tracking
Without temporal averaging, such as rate codes, it remains challenging to train spiking neural networks for temporal regression tasks. In this work, we present a novel method to accurately predict spatial coordinates from event data with a fully spiking convolutional neural network (SCNN) without temporal averaging. Our method performs on-par with artificial neural networks (ANN) of similar complexity. Additionally, we demonstrate faster convergence in half the time using translation- and scale-invariant receptive fields. To permit comparison with conventional frame-based ANNs, we base our results on a simulated event-based dataset with an unrealistic high density. Therefore, we hypothesize that our method significantly outperform ANNs in settings with lower event density, as seen in real-life event-based data. Our model is fully spiking and can be ported directly to neuromorphic hardware.