Section
Computer Science
Sparse Attention Mechanisms for Low-Resource Language Models
A learned sparse-attention scheme cuts inference cost for morphologically rich, low-resource languages with no measurable loss in accuracy.
Ananya Rao, Devashish Kumar · Vol 1, Issue 1 · pp. 1-11
Differentially Private Recommendation at the Edge
A federated recommender adds calibrated noise on-device, preserving utility under a strict privacy budget.
Ananya Rao · Vol 1, Issue 1 · pp. 21-28
On-Device Speech Recognition for Indian Languages
A compact acoustic model runs fully on-device across nine Indian languages with competitive word-error rates.
Karthik Nair · Vol 1, Issue 1 · pp. 12-20
Quantised Transformers for Embedded Vision
Post-training quantisation retains accuracy on embedded vision tasks at a quarter of the memory footprint.
Ananya Rao · Vol 1, Issue 1 · pp. 29-40