On Interpretability of Linear Autoencoders
The result's identifiers
Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F24%3A10494534" target="_blank" >RIV/00216208:11320/24:10494534 - isvavai.cz</a>
Alternative codes found
RIV/68407700:21340/24:00381456
Result on the web
<a href="https://doi.org/10.1145/3640457.3688179" target="_blank" >https://doi.org/10.1145/3640457.3688179</a>
DOI - Digital Object Identifier
<a href="http://dx.doi.org/10.1145/3640457.3688179" target="_blank" >10.1145/3640457.3688179</a>
Alternative languages
Result language
angličtina
Original language name
On Interpretability of Linear Autoencoders
Original language description
We derive a novel graph-based interpretation of linear autoencoder models easer, slim, and their approximate variants. Contrary to popular belief, we reveal that the weights of these models should not be interpreted as dichotomic item similarity but merely as its magnitude. Consequently, we propose a simple modification that considerably improves retrieval ability in sparse domains and yields interpretable inference with negative inputs, as demonstrated by both offline and online experiments. Experiment codes and extended results are available at https://osf.io/bjmuv/.
Czech name
—
Czech description
—
Classification
Type
D - Article in proceedings
CEP classification
—
OECD FORD branch
10201 - Computer sciences, information science, bioinformathics (hardware development to be 2.2, social aspect to be 5.8)
Result continuities
Project
<a href="/en/project/GA22-21696S" target="_blank" >GA22-21696S: Deep Visual Representations of Unstructured Data</a><br>
Continuities
P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)
Others
Publication year
2024
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů
Data specific for result type
Article name in the collection
PROCEEDINGS OF THE EIGHTEENTH ACM CONFERENCE ON RECOMMENDER SYSTEMS, RECSYS 2024
ISBN
979-8-4007-0505-2
ISSN
—
e-ISSN
—
Number of pages
6
Pages from-to
975-980
Publisher name
ASSOC COMPUTING MACHINERY
Place of publication
NEW YORK
Event location
Bari
Event date
Oct 14, 2024
Type of event by nationality
WRD - Celosvětová akce
UT code for WoS article
001336908500129