Efficient Representations for Large Dynamic Sequences in ML

Arthur Charguéraud 1 Mike Rainey 2
1 CAMUS - Compilation pour les Architectures MUlti-coeurS
Inria Nancy - Grand Est, ICube - Laboratoire des sciences de l'ingénieur, de l'informatique et de l'imagerie
Abstract : The use of sequence containers, including stacks, queues, and double-ended queues, is ubiquitous in programming. When the maximal number of elements is not known in advance, containers need to grow dynamically. For this purpose, most ML programs either rely on lists or vectors. These structures are inefficient, both in terms of time and space usage. We investigate the use of chunked-based data structures. Such structures save a lot of memory and may deliver better performance than classic container data structures. We observe a 2x speedup compared with vectors, and up to a 3x speedup compared with lengthy lists.
Document type :
Poster communications
Complete list of metadatas

Cited literature [7 references]  Display  Hide  Download

https://hal.inria.fr/hal-01669407
Contributor : Arthur Charguéraud <>
Submitted on : Wednesday, December 20, 2017 - 8:20:12 PM
Last modification on : Monday, November 26, 2018 - 1:28:01 PM

File

chunkseq_ml.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-01669407, version 1

Citation

Arthur Charguéraud, Mike Rainey. Efficient Representations for Large Dynamic Sequences in ML. ML Family Workshop, Sep 2017, Oxford, United Kingdom. 2017. ⟨hal-01669407⟩

Share

Metrics

Record views

204

Files downloads

72