Dr. Cristina Luna Jiménez

Wissenschaftlice Mitarbeiterin
Lehrstuhl für Menschzentrierte Künstliche Intelligenz
Telefon: +49 821 598 2348
E-Mail:
Raum: 2015 (N)
Sprechzeiten: Nach Vereinbarung
Adresse: Universitätsstraße 6a, 86159 Augsburg

Research Interests

  • Deep Learning
  • Affective-Computing (e.g. Multimodal Emotion Recognition, Trustworthiness Recognition)
  • Sign Language Recognition and Translation
  • Generative AI (Large Language Models)
  • Natural Language Processing
  • Signal Processing
(WEB UNDER DEVELOPMENT, FOR MORE INFORMATION CHECK THE OPEN THESIS TOPICS AND THE LINKS)
 
 

Links

Academic Activities

  • Reviewer: 
    • Review activities for Northern Lights Deep Learning Conference 2023
    • Review activities for Journal of Big Data (Springer Nature)
    • Review activities for Neural Processing Letters (Springer Nature)
    • Review activities for SIGDIAL Conference 2024
  • Memberships:
    • Member of RTTH (Red Temática de Tecnologías del Habla)
  • Research Stays: 
    • 3-months in the University Sains Malaysia
    • 2-months in GLOBIT as part of the MENHIR project

 

 

 

 

 

Awards

  • National Price ‘El futuro de las TIC - Beca 2023’of Huawei for the talent of STEAM (Science-Technology-Engineering-Art-Mathematics) women. [5.000 €, Mar 2023]
  • 3rd position - Best student journal paper of the Red Temática en Tecnologías del Habla - Edition 2022, with the article: “Multimodal Emotion Recognition on RAVDESS Dataset Using Transfer Learning”.
  • Finalist of the U.P.M. challenge of `Three Minute Thesis’, which follows the guidelines of the Queensland University (Australia) [200 €, May 2023]
 
  • Grant for trips and presentations in Congress[600 €, May 2023]
  • Research support Grant of the E.T.S.I.T. – U.P.M. for IberSPEECH 2022 congress presentations [400 €, Jan 2023]
  • Mobility Grant from the Programa Propio - Universidad Politécnica de Madrid, for a 3-month research stay at the Universiti Sains Malaysia (Penang, Malaysia) [4.817,03 €, June 2022]
  • Erasmus KA107 for the 3-month research stay at the Universiti Sains Malaysia [3.763€, January 2018]
  • Erasmus+ for the 1-year of a Msc.-exchange program at the KU Leuven [1.799€, July 2017]
  • Grant for studying Bsc. programme from the ‘Patronato Ángel Barbero Martín de Vidales’ [9.000€, Sept 2014]
 
 

Projects

Supervised Theses

  • Design and implementation of computational models employing deep-neural networks for performing emotion recognition in multimedia content. (Msc. 2022, Universidad Politécnica de Madrid) 
  • Design and implementation of computational models for performing deception detection (Bsc. 2022, Universidad Autónoma de Madrid)
  • Design and implementation of computational models for predicting rhetoric skills (Bsc. 2021, Universodad Politécnica de Madrid)
  • Design and implementation of computational models for transforming facial expressions in videos (Msc. 2021, Universodad Politécnica de Madrid)
  • Design and implementation of computational models for transforming facial expressions in images (Bsc. 2020, Universodad Politécnica de Madrid)
  • Design and implementation of computational models for film assessment and classification (Msc. 2020, Universodad Politécnica de Madrid)

Open Thesis Topics

The following topics can be flexibly varied in scope and orientation, so that the realization as a bachelor thesis, master thesis or project module is possible. Furthermore, the focus of the content can of course be aligned with the interests of the student.

Furthermore, I am always happy to receive your own suggestions for topics, as long as they show a certain overlap with my research focus.

 

 

Emotion & Trustworthiness Recognition

Affective Computing has its origin on 1997 as a branch of Artificial Intelligence that had as aim solving the question of wether computers could be able to recognize, interpret, process and even simulate human emotions. Nowadays, it is still an open and active research topic that has been extended to other complex cognitive states (e.g. engagement, trustworthiness, etc.). In this line, we offer several Bsc. and Msc. theses to perform recognition (or generation) of these mental-congitive states by employing transformers and derived state-of-the-art architectures. 

Sign Language Recognition

According to the WHO, over 1.5 billion people experience some degree of hearing loss. In this scenario, being able to develop technilogy to asist deaf people is a critical issue to let them be integrated into society. In this line, we offer several projects oriented to perform Sign Language Recognition usign contrastive learning, self-supervised and supervised learning paradigms. 

 

Lip Reading

Lip Reading is a branch of study that aims to identify what it is said by focusing uniquely in the facial movements (normally of the mouth region). This project has the aim of developing models able to recognize the pronounced words from the images of a video. This challenging task has multiple benefits such as the employment of this models for improving the recognition of the transmited messages in Speech-to-Text conversions or in Sign Language Recognition. We will explore this use-case training Transformers and Masked Auto-Encoders. 

Sign Language Alignment

In this project, we will dive into the process of alignment transcriptions with their equivalent words in Sign Language (called glosses) under scarce resources scenarios. In order to achieve this aim, we will explore several methods with  pre-trained models on similar tasks (e.g. activity recognition) to detect when a specific gloss appears in the video by calculating similarities between the actual video and the predicions or their extracted vectors. 

Facial Expressions and Sign Language Recognition

In this project we will explore multi-task learning set-ups by employing the same machine-learning model to learn two task at the same time, in our case, facial expressions and sign language from videos of Sign Language signers. 

Large Language Models as translators

Since the appartion of ChatGPT, Large Language Models (LLMs) have received attention from the public and the research community. In this project, we will explore their capabilities for understanding Sign Language.

Large Language Models as detectors

Since the appartion of ChatGPT, Large Language Models (LLMs) have received attention from the public and the research community. In this project, we will explore their capabilities of Large Language Models to process text and perform classification tasks. Among the possibilities, we offer: topic classification, mental-health disorder detection, emotion or valence recognition, etc. For performing this techniques we will follow several approaches, from Zero-Shot applying Prompt Engineering techniques, Embeddings Extraction, or Fine-Tuning with LORA-derived methods. 

Lehre

(Angewandte Filter: Semester: aktuelles | Dozenten: Cristina Luna Jiménez | Vorlesungsarten: alle)
Name Semester Typ
Übung zu Generative AI for Human-Computer Interaction Lab Wintersemester 2024/25 Übung
Generative AI for Human-Computer Interaction Lab Wintersemester 2024/25 Vorlesung

Suche