Skip to Main content Skip to Navigation
Conference papers

How do People Train a Machine? Strategies and (Mis)Understandings

Téo Sanchez 1 Baptiste Caramiaux 2, 1 Jules Françoise 3, 4, 5, 6 Frédéric Bevilacqua 7 Wendy Mackay 1
1 EX-SITU - Extreme Situated Interaction
Inria Saclay - Ile de France, LISN - Laboratoire Interdisciplinaire des Sciences du Numérique, IaH - Interaction avec l'Humain
5 AMI - Architecture et Modèles pour l'Interaction
LIMSI - Laboratoire d'Informatique pour la Mécanique et les Sciences de l'Ingénieur
6 AMI - Architectures et Modèles pour l'Interaction
LISN - Laboratoire Interdisciplinaire des Sciences du Numérique, IaH - Interaction avec l'Humain
Abstract : Machine learning systems became pervasive in modern interactive technology but provide users with little, if any, agency with respect to how their models are trained from data. In this paper, we are interested in the way novices handle learning algorithms, what they understand from their behavior and what strategy they may use to "make it work". We developed a web-based sketch recognition algorithm based on Deep Neural Network (DNN), called Marcelle-Sketch, that end-users can train incrementally. We present an experimental study that investigate people's strategies and (mis)understandings in a realistic algorithm-teaching task. Our study involved 12 participants who performed individual teaching sessions using a think-aloud protocol. Our results show that participants adopted heterogeneous strategies in which variability affected the model performances. We highlighted the importance of sketch sequencing, particularly at the early stage of the teaching task. We also found that users' understanding is facilitated by simple operations on drawings, while confusions are caused by certain inherent properties of DNN. From these findings, we propose implications for design of IML systems dedicated to novices and discuss the socio-cultural aspect of this research.
Complete list of metadata

https://hal.inria.fr/hal-03182950
Contributor : Téo Sanchez <>
Submitted on : Friday, March 26, 2021 - 6:05:12 PM
Last modification on : Tuesday, July 13, 2021 - 2:17:13 PM
Long-term archiving on: : Sunday, June 27, 2021 - 7:12:08 PM

File

V5cscw162-sanchezA.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-03182950, version 1

Citation

Téo Sanchez, Baptiste Caramiaux, Jules Françoise, Frédéric Bevilacqua, Wendy Mackay. How do People Train a Machine? Strategies and (Mis)Understandings. CSCW 2021 - The 24th ACM Conference on Computer-Supported Cooperative Work and Social Computing, Oct 2021, Virtual, United States. ⟨hal-03182950⟩

Share

Metrics

Record views

216

Files downloads

676