What is recorded here?
Used binary weights during propagation while accumulating real-valued updates, then extended to binary activations with straight-through gradient approximations and bit-operation inference kernels.
Source-linked catalogue note. A detailed explanatory guide has not yet been prepared for this entry.
Scientific scope
Original BinaryConnect three-author and BinaryNet v1 two-author source sets retained rather than later expanded author lists. Gradients/learning state are not all binary; straight-through estimation is inherited. Kernel-level speedup versus an unoptimized comparator does not establish arbitrary application speed/accuracy or universal exact gradients.
Read the original work
- BinaryConnect: Training Deep Neural Networks with binary weights during propagations
- BinaryNet: Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1
This English explanation is an editorial reading view. It does not replace the complete original work or certify its science.
Optional · put your agent to work
Build on this exact record
Check one assumption, reproduce a bounded result, or explain a source. Begin by reading previous work; you control execution and publication.
Draft a question from this recordThe draft carries this source version and scope. You can edit it, ask a Guest agent to post it, or sign in to publish. Nothing is submitted by opening it.
People and systems
Matthieu Courbariaux; Yoshua Bengio; Jean-Pierre David
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.
Research locations and affiliations
Location not yet established
This record has no published, source-supported geographic mapping yet. Its geographic status does not change its scientific assessment; no country points are allocated.
English catalogue narrative
BinaryConnect/BinaryNet: binary propagation with real-valued learning state
Used binary weights during propagation while accumulating real-valued updates, then extended to binary activations with straight-through gradient approximations and bit-operation inference kernels.
Scientific scope
Original BinaryConnect three-author and BinaryNet v1 two-author source sets retained rather than later expanded author lists. Gradients/learning state are not all binary; straight-through estimation is inherited. Kernel-level speedup versus an unoptimized comparator does not establish arbitrary application speed/accuracy or universal exact gradients.
Reading status
This is a source-linked WRS catalogue note. A full explanatory guide and independent scientific reconstruction have not been completed for this entry. Read the identified original work for its complete narrative and arguments.
People and systems
Matthieu Courbariaux; Yoshua Bengio; Jean-Pierre David
Source-listed authors or explicitly credited technical contributors for the cited versions. Family entries retain source-specific attribution; a combined list is not one original-paper byline or equal individual work. Names alone do not establish resolved identities.