Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

Does small talk with a medical provider affect ChatGPT's medical counsel? Performance of ChatGPT on USMLE with and without distractions.

PloS one·2024
Same author

Conversational Agents: Goals, Technologies, Vision and Challenges.

Sensors (Basel, Switzerland)·2021
Same author

A Safe Collaborative Chatbot for Smart Home Assistants.

Sensors (Basel, Switzerland)·2021
Same author

Vision-Less Sensing for Autonomous Micro-Drones.

Sensors (Basel, Switzerland)·2021
Same author

Reactive stroma and trastuzumab resistance in HER2-positive early breast cancer.

International journal of cancer·2020

Related Experiment Video

Updated: Dec 7, 2025

Modification of a Colliculo-thalamocortical Mouse Brain Slice, Incorporating 3-D printing of Chamber Components and Multi-scale Optical Imaging
06:05

Modification of a Colliculo-thalamocortical Mouse Brain Slice, Incorporating 3-D printing of Chamber Components and Multi-scale Optical Imaging

Published on: September 18, 2015

8.6K

SAIF: A Correction-Detection Deep-Learning Architecture for Personal Assistants.

Amos Azaria1, Keren Nivasch1

  • 1Data Science Center, Ariel University, Ariel 40700, Israel.

Sensors (Basel, Switzerland)
|October 2, 2020
PubMed
Summary
This summary is machine-generated.

Intelligent agents can now detect user corrections using deep learning. The SAIF model analyzes voice and text to identify and fix command errors, improving agent performance.

Keywords:
correction detectiondeep learninghuman–agent interactionimplicit feedbackmultimodal architecture

More Related Videos

An Open-Source Normothermic Perfusion System Designed for Research Scientists
11:23

An Open-Source Normothermic Perfusion System Designed for Research Scientists

Published on: July 18, 2025

833
A Novel Method for Localizing Reporter Fluorescent Beads Near the Cell Culture Surface for Traction Force Microscopy
13:30

A Novel Method for Localizing Reporter Fluorescent Beads Near the Cell Culture Surface for Traction Force Microscopy

Published on: September 16, 2014

14.9K

Related Experiment Videos

Last Updated: Dec 7, 2025

Modification of a Colliculo-thalamocortical Mouse Brain Slice, Incorporating 3-D printing of Chamber Components and Multi-scale Optical Imaging
06:05

Modification of a Colliculo-thalamocortical Mouse Brain Slice, Incorporating 3-D printing of Chamber Components and Multi-scale Optical Imaging

Published on: September 18, 2015

8.6K
An Open-Source Normothermic Perfusion System Designed for Research Scientists
11:23

An Open-Source Normothermic Perfusion System Designed for Research Scientists

Published on: July 18, 2025

833
A Novel Method for Localizing Reporter Fluorescent Beads Near the Cell Culture Surface for Traction Force Microscopy
13:30

A Novel Method for Localizing Reporter Fluorescent Beads Near the Cell Culture Surface for Traction Force Microscopy

Published on: September 16, 2014

14.9K

Area of Science:

  • Artificial Intelligence
  • Human-Computer Interaction
  • Machine Learning

Background:

  • Intelligent agents increasingly use natural language, but misinterpretations of user commands are common.
  • Users often rephrase commands to correct agent errors, indicating a need for systems to recognize these corrections.
  • Current systems lack robust mechanisms for automatically detecting and learning from user-initiated command corrections.

Purpose of the Study:

  • To develop a deep learning-based method for automatically detecting user corrections of intelligent agent commands.
  • To enhance intelligent agent performance by enabling them to recognize and adapt to user corrections.
  • To introduce a multimodal architecture that leverages both voice and textual command data.

Main Methods:

  • Developed a multimodal deep learning architecture named SAIF (Speech, Audio, and Intent Fusion).
  • SAIF processes user voice commands (including tone, speed, emphasis) and their transcripts.
  • The model integrates agent feedback on command execution success/failure with multimodal command data.

Main Results:

  • SAIF effectively detects user corrections by analyzing voice cues and command transcripts.
  • The model demonstrated superior performance compared to existing state-of-the-art methods on a newly released dataset.
  • The multimodal approach, incorporating audio and text, proved crucial for accurate correction detection.

Conclusions:

  • Automatic detection of user corrections is feasible and beneficial for intelligent agents.
  • The SAIF architecture offers a promising approach for improving human-agent interaction by reducing errors.
  • Further research can build upon this multimodal strategy to create more robust and adaptive intelligent systems.