Nitin Chandrachoodan

Shashank Nag, Gourav Datta, Souvik Kundu et al.

Vision Transformer models, such as ViT, Swin Transformer, and Transformer-in-Transformer, have recently gained significant traction in computer vision tasks due to their ability to capture the global relation between features which leads to superior performance. However, they are compute-heavy and difficult to deploy in resource-constrained edge devices. Existing hardware accelerators, including those for the closely-related BERT transformer models, do not target highly resource-constrained environments. In this paper, we address this gap and propose ViTA - a configurable hardware accelerator for inference of vision transformer models, targeting resource-constrained edge computing devices and avoiding repeated off-chip memory accesses. We employ a head-level pipeline and inter-layer MLP optimizations, and can support several commonly used vision transformer models with changes solely in our control logic. We achieve nearly 90% hardware utilization efficiency on most vision transformer models, report a power of 0.88W when synthesised with a clock of 150 MHz, and get reasonable frame rates - all of which makes ViTA suitable for edge applications.

Swathi Shree Narashiman, Nitin Chandrachoodan

Data compression continues to evolve, with traditional information theory methods being widely used for compressing text, images, and videos. Recently, there has been growing interest in leveraging Generative AI for predictive compression techniques. This paper introduces a lossless text compression approach using a Large Language Model (LLM). The method involves two key steps: first, prediction using a dense neural network architecture, such as a transformer block; second, compressing the predicted ranks with standard compression algorithms like Adaptive Huffman, LZ77, or Gzip. Extensive analysis and benchmarking against conventional information-theoretic baselines demonstrate that neural compression offers improved performance.

Abinand Nallathambi, Nitin Chandrachoodan

Evaluation of spiking neural networks requires fetching a large number of synaptic weights to update postsynaptic neurons. This limits parallelism and becomes a bottleneck for hardware. We present an approach for spike propagation based on a probabilistic interpretation of weights, thus reducing memory accesses and updates. We study the effects of introducing randomness into the spike processing, and show on benchmark networks that this can be done with minimal impact on the recognition accuracy. We present an architecture and the trade-offs in accuracy on fully connected and convolutional networks for the MNIST and CIFAR10 datasets on the Xilinx Zynq platform.

Gargi Mitra, Prasanna Karthik Vairam, Patanjali SLPSK et al.

Privacy leaks from Netflix videos/movies is well researched. Current state-of-the-art works have been able to obtain coarse-grained information such as the genre and the title of videos by passive observation of encrypted traffic. However, leakage of fine-grained information from encrypted traffic has not been studied so far. Such information can be used to build behavioural profiles of viewers. On 28th December 2018, Netflix released the first mainstream interactive movie called 'Black Mirror: Bandersnatch'. In this work, we use this movie as a case-study to show for the first time that fine-grained information (i.e., choices made by users) can be revealed from encrypted traffic. We use the state information exchanged between the viewer's browser and Netflix as the side-channel. To evaluate our proposed technique, we built the first interactive video traffic dataset of 100 viewers; which we will be releasing. Preliminary results indicate that the choices made by a user can be revealed 96% of the time in the worst case.