David Semedo
Ph.D. Researcher at NOVA University of Lisbon

About Me

I'm a Researcher at the NOVA Laboratory for Computer Science and Informatics (NOVA LINCS), from NOVA School of Science and Technology of NOVA University of Lisbon.

My work focuses on bridging vision and language for multimedia understanding . I research deep learning approaches to address the semantic gap between visual and textual modalities. Currently working on Deep learning models for Multimedia, Multimodal and Cross-modal embeddings, and Visual Dialog systems.

Expertise and Interests: Multimedia Understanding. Multimodal machine learning. Neural-based Representation Learning. Computer vision and natural language processing methods for video, image and text modelling. My research interests lye around the field of multimedia understanding, namely in the intersection of computer vision and NLP, Deep Neural Networks, Artificial Intelligence and Data Mining.

Current positions: If you are interested in working on the fields of deep learning for multimedia, vision and language understanding (visual question-answering, image-captioning, etc.) or neural representation learning, we are currently looking for motivated students.

Academic Degrees

2016-2020 Ph.D. in Deep Learning for Multimedia, by NOVA School of Science and Technology, under the supervision of Prof. João Magalhães
Thesis: Bridging Vision and Language over Time with Neural Cross-modal Embeddings. [Thesis] [Bib]

2015 M.Sc. in Computer Science by NOVA School of Science and Technology, under the supervision of Prof. Pedro Barahona and Prof. Pedro Medeiros.
Thesis: A Public Transport Bus Assignment Problem: Parallel Metaheuristics Assessment. [Thesis] [Bib]

2013 B.Sc. in Computer Science by NOVA School of Science and Technology.

Publications

Adaptive Temporal Triplet-loss for Cross-modal Embedding Learning
David Semedo, and João Magalhães. ACM Multimedia 2020, Seattle, USA, Oct. 2020.
[Project Page (w/ Dataset)] [PDF (To Appear)] [Bib]

Diachronic Cross-modal Embeddings
David Semedo, and João Magalhães. ACM Multimedia 2019, Nice, France, Oct. 2019.
[Project Page (w/ Dataset)] [PDF] [Bib]

Cross-Modal Subspace Learning with Scheduled Adaptive Margin Constraints
David Semedo, and João Magalhães. ACM Multimedia 2019, Nice, France, Oct. 2019.
[PDF] [Bib]

A Benchmark of Visual Storytelling in Social Media
Gonçalo Marcelino, David Semedo, André Mourão, Saverio Blasi, Marta Mrak, and João Magalhães. International Conference on Multimedia Retrieval 2019, Ottawa, Canada, June 2019.
[PDF] [Bib]

Dynamic-Keyword Extraction from Social Media
David Semedo, and João Magalhães. European Conference on Information Retrieval 2019, Cologne, Germany, April 2019.
[PDF] [Bib]

Temporal Cross-Media Retrieval with Soft-Smoothing
David Semedo, and João Magalhães. ACM Multimedia 2018, Seoul, Korea, Oct. 2018.
[PDF] [Bib]

TRECVID 2018: Benchmarking video activity detection, video captioning and matching, video storytelling linking and video search
Awad, George, Asad Butt, Keith Curtis, Yooyoung Lee, Jonathan Fiscus, Afzad Godil, David Joy, Andrew Delgado, Alan Smeaton, Yvette Graham, et al.. Proceedings of TRECVID 2018, 2018
[PDF] [Bib]

TRECVID 2018 Linking Task: Social Media Video Storytelling
João Magalhães, David Semedo, Saverio Blasi. TRECVID 2018.
[PDF]

Inferring User Gender from User Generated Visual Content on a Deep Semantic Space
David Semedo, Flávio Martins, and João Magalhães. 26th European Signal Processing Conference (EUSIPCO 2018), Rome, Italy, 2018.
[PDF] [Bib]

NovaSearch at ImageCLEFmed 2016 Subfigure Classification Task
David Semedo, João Magalhães. Working Notes of CLEF 2016 - Conference and Labs of the Evaluation forum, Évora, Portugal, Sep. 2016.
[PDF] [Bib]

Asynchronous Parallel Ant-Colony Optimization Strategies: Application to the Multi-Depot Vehicle Scheduling Problem with Line Exchanges
David Semedo, Pedro Barahona, Pedro Medeiros. 7th INForum, Covilhã, Portugal, Sep. 2015.
[PDF] [Bib]

Activities

Please find below a list of my most relevant academic activities, grouped by year.

2019:

  • Seminar Talk: Bridging Vision and Language over Time with Neural Cross-modal Embeddings. [Details]

  • PC Member for ACM Multimedia 2019 (MM 2019).

2018:

  • Reviewer for ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR 2018).

  • Reviewer for ACM International Conference on Multimedia Retrieval 2018 (ICMR 2018).

  • Reviewer for International Conference on Theory and Practice of Digital Libraries 2018 (TPDL 2018).

2017:

  • Reviewer for ACM International Conference on Multimedia Retrieval 2017 (ICMR 2017).

  • Reviewer for IEEE International Conference on Multimedia and Expo 2017 (ICME 2017).

2016:

  • Reviewer for ACM International Conference on Multimedia Retrieval 2016 (ICMR 2016).

  • Monitor at the LxMLS - 2016 Lisbon Machine Learning Summer School.

  • Poster Presentation at CLEF 2016 - Conference and Labs of the Evaluation Forum.

Teaching

2020 Computational Logic - B.Sc. Computer Science course.

2019 Algorithms and Data Structures - B.Sc. Computer Science course.

2018 Web Search and Data Mining - M.Sc. Computer Science course. [Webpage]

2016 Lab Instructor at LxMLS - 2016 Lisbon Machine Learning Summer School. [Webpage]

2015 Algorithms and Data Structures - B.Sc. Computer Science course.

2014 Algorithms and Data Structures - B.Sc. Electrical and Computer Engineering course.

Projects and Others

2018-Present - Visual Stories: Cross-modal Image and Video Search/Browsing Tool for Social Media Event Data (Ongoing Work) [Link]

2016 - Classical and Neural Language Models Survey [PDF]
Statistical Learning Course final project, lectured by Prof. Mário Figueiredo