Graph Expansion in Pruned Recurrent Neural Network Layers Preserve Performance

no code implementations • 17 Mar 2024 • Suryam Arnav Kalra, Arindam Biswas, Pabitra Mitra, BISWAJIT BASU

We prune recurrent networks such as RNNs and LSTMs, maintaining a large spectral gap of the underlying graphs and ensuring their layerwise expansion properties.

Revisiting the Lottery Ticket Hypothesis: A Ramanujan Graph Perspective

no code implementations • 29 Sep 2021 • BITHIKA PAL, Arindam Biswas, Pabitra Mitra, BISWAJIT BASU

Subsequently, with the loss of Ramanujan graph property, accuracy begins to reduce sharply.