PROM quantizes pointwise convolutions to ternary weights and all other layers to 8 bits, converting the dominant CNN operations to additions and yielding order-of-magnitude estimated energy savings at similar accuracy.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
PROM: Prioritize Reduction of Multiplications Over Lower Bit-Widths for Efficient CNNs
PROM quantizes pointwise convolutions to ternary weights and all other layers to 8 bits, converting the dominant CNN operations to additions and yielding order-of-magnitude estimated energy savings at similar accuracy.