Marah Halawa
Photo: SCIoI
Marah Halawa is a computer scientist specializing in machine learning and artificial intelligence, with experience applying these methods to areas including speech recognition, facial expression recognition, and fraud detection.
She joined SCIoI as a doctoral researcher in October 2020.
Marah received her M.Sc. in Computer Science from TU Berlin in 2019, where she specialized in Machine Learning and Artificial Intelligence and worked on many machine learning applications. After graduation, she started a full-time position as a Data Scientist, where she developed a deep-learning-based system for fraud detection.
At SCIoI, Marah worked on Project 29. Her research focused on enabling visual understanding in modularized and hierarchical temporal vision systems for closed perception-action loops.
Projects
Marah Halawa is member of:
Publications +
6984777
Halawa
1
apa
50
date
desc
year
19862
https://www.scienceofintelligence.de/wp-content/plugins/zotpress/
%7B%22status%22%3A%22success%22%2C%22updateneeded%22%3Afalse%2C%22instance%22%3Afalse%2C%22meta%22%3A%7B%22request_last%22%3A0%2C%22request_next%22%3A0%2C%22used_cache%22%3Atrue%7D%2C%22data%22%3A%5B%7B%22key%22%3A%22QKAHMER8%22%2C%22library%22%3A%7B%22id%22%3A6984777%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Halawa%20et%20al.%22%2C%22parsedDate%22%3A%222024-06%22%2C%22numChildren%22%3A1%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20padding-left%3A%201em%3B%20text-indent%3A-1em%3B%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%26gt%3BHalawa%2C%20M.%2C%20Blume%2C%20F.%2C%20Bideau%2C%20P.%2C%20Maier%2C%20M.%2C%20Rahman%2C%20R.%20A.%2C%20%26amp%3B%20Hellwich%2C%20O.%20%282024%29.%20Multi-Task%20Multi-Modal%20Self-Supervised%20Learning%20for%20Facial%20Expression%20Recognition.%20%26lt%3Bi%26gt%3B2024%20IEEE%5C%2FCVF%20Conference%20on%20Computer%20Vision%20and%20Pattern%20Recognition%20Workshops%20%28CVPRW%29%26lt%3B%5C%2Fi%26gt%3B%2C%204604%26%23x2013%3B4614.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1109%5C%2FCVPRW63382.2024.00463%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1109%5C%2FCVPRW63382.2024.00463%26lt%3B%5C%2Fa%26gt%3B%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Multi-Task%20Multi-Modal%20Self-Supervised%20Learning%20for%20Facial%20Expression%20Recognition%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Marah%22%2C%22lastName%22%3A%22Halawa%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Florian%22%2C%22lastName%22%3A%22Blume%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Pia%22%2C%22lastName%22%3A%22Bideau%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Martin%22%2C%22lastName%22%3A%22Maier%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Rasha%20Abdel%22%2C%22lastName%22%3A%22Rahman%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Olaf%22%2C%22lastName%22%3A%22Hellwich%22%7D%5D%2C%22abstractNote%22%3A%22Human%20communication%20is%20multi-modal%3B%20e.g.%2C%20face-to-face%20interaction%20involves%20auditory%20signals%20%28speech%29%20and%20visual%20signals%20%28face%20movements%20and%20hand%20gestures%29.%20Hence%2C%20it%20is%20essential%20to%20exploit%20multiple%20modalities%20when%20designing%20machine%20learning-based%20facial%20expression%20recognition%20systems.%20In%20addition%2C%20given%20the%20ever-growing%20quantities%20of%20video%20data%20that%20capture%20human%20facial%20expressions%2C%20such%20systems%20should%20utilize%20raw%20unlabeled%20videos%20without%20requiring%20expensive%20annotations.%20Therefore%2C%20in%20this%20work%2C%20we%20employ%20a%20multitask%20multi-modal%20self-supervised%20learning%20method%20for%20facial%20expression%20recognition%20from%20in-the-wild%20video%20data.%20Our%20model%20combines%20three%20self-supervised%20objective%20functions%3A%20First%2C%20a%20multi-modal%20contrastive%20loss%2C%20that%20pulls%20diverse%20data%20modalities%20of%20the%20same%20video%20together%20in%20the%20representation%20space.%20Second%2C%20a%20multi-modal%20clustering%20loss%20that%20preserves%20the%20semantic%20structure%20of%20input%20data%20in%20the%20representation%20space.%20Finally%2C%20a%20multi-modal%20data%20reconstruction%20loss.%20We%20conduct%20a%20comprehensive%20study%20on%20this%20multimodal%20multi-task%20self-supervised%20learning%20method%20on%20three%20facial%20expression%20recognition%20benchmarks.%20To%20that%20end%2C%20we%20examine%20the%20performance%20of%20learning%20through%20different%20combinations%20of%20self-supervised%20tasks%20on%20the%20facial%20expression%20recognition%20downstream%20task.%20Our%20model%20ConCluGen%20outperforms%20several%20multi-modal%20self-supervised%20and%20fully%20supervised%20baselines%20on%20the%20CMU-MOSEI%20dataset.%20Our%20results%20generally%20show%20that%20multi-modal%20self-supervision%20tasks%20offer%20large%20performance%20gains%20for%20challenging%20tasks%20such%20as%20facial%20expression%20recognition%2C%20while%20also%20reducing%20the%20amount%20of%20manual%20annotations%20required.%20We%20release%20our%20pre-trained%20models%20as%20well%20as%20source%20code%20publicly1.%22%2C%22proceedingsTitle%22%3A%222024%20IEEE%5C%2FCVF%20Conference%20on%20Computer%20Vision%20and%20Pattern%20Recognition%20Workshops%20%28CVPRW%29%22%2C%22conferenceName%22%3A%222024%20IEEE%5C%2FCVF%20Conference%20on%20Computer%20Vision%20and%20Pattern%20Recognition%20Workshops%20%28CVPRW%29%22%2C%22date%22%3A%222024-06%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%2210.1109%5C%2FCVPRW63382.2024.00463%22%2C%22ISBN%22%3A%22%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fieeexplore.ieee.org%5C%2Fdocument%5C%2F10678641%22%2C%22ISSN%22%3A%222160-7516%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%5D%2C%22dateModified%22%3A%222026-06-22T12%3A01%3A10Z%22%7D%7D%2C%7B%22key%22%3A%22DCZ2RZ22%22%2C%22library%22%3A%7B%22id%22%3A6984777%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Halawa%20et%20al.%22%2C%22parsedDate%22%3A%222022%22%2C%22numChildren%22%3A1%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20padding-left%3A%201em%3B%20text-indent%3A-1em%3B%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%26gt%3BHalawa%2C%20M.%2C%20Hellwich%2C%20O.%2C%20%26amp%3B%20Bideau%2C%20P.%20%282022%29.%20Action-Based%20Contrastive%20Learning%20for%20Trajectory%20Prediction.%20In%20S.%20Avidan%2C%20G.%20Brostow%2C%20M.%20Ciss%26%23xE9%3B%2C%20G.%20M.%20Farinella%2C%20%26amp%3B%20T.%20Hassner%20%28Eds.%29%2C%20%26lt%3Bi%26gt%3BComputer%20Vision%20%26%23x2013%3B%20ECCV%202022%26lt%3B%5C%2Fi%26gt%3B%20%28Vol.%2013699%2C%20pp.%20143%26%23x2013%3B159%29.%20Springer%20Nature%20Switzerland.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1007%5C%2F978-3-031-19842-7_9%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.1007%5C%2F978-3-031-19842-7_9%26lt%3B%5C%2Fa%26gt%3B%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22conferencePaper%22%2C%22title%22%3A%22Action-Based%20Contrastive%20Learning%20for%20Trajectory%20Prediction%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Marah%22%2C%22lastName%22%3A%22Halawa%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Olaf%22%2C%22lastName%22%3A%22Hellwich%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Pia%22%2C%22lastName%22%3A%22Bideau%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Shai%22%2C%22lastName%22%3A%22Avidan%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Gabriel%22%2C%22lastName%22%3A%22Brostow%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Moustapha%22%2C%22lastName%22%3A%22Ciss%5Cu00e9%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Giovanni%20Maria%22%2C%22lastName%22%3A%22Farinella%22%7D%2C%7B%22creatorType%22%3A%22editor%22%2C%22firstName%22%3A%22Tal%22%2C%22lastName%22%3A%22Hassner%22%7D%5D%2C%22abstractNote%22%3A%22%22%2C%22proceedingsTitle%22%3A%22Computer%20Vision%20%5Cu2013%20ECCV%202022%22%2C%22conferenceName%22%3A%22%22%2C%22date%22%3A%222022%22%2C%22eventPlace%22%3A%22%22%2C%22DOI%22%3A%2210.1007%5C%2F978-3-031-19842-7_9%22%2C%22ISBN%22%3A%22978-3-031-19841-0%20978-3-031-19842-7%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Flink.springer.com%5C%2F10.1007%5C%2F978-3-031-19842-7_9%22%2C%22ISSN%22%3A%22%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%5D%2C%22dateModified%22%3A%222026-06-22T11%3A50%3A40Z%22%7D%7D%2C%7B%22key%22%3A%224YNQ2UHV%22%2C%22library%22%3A%7B%22id%22%3A6984777%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Halawa%20et%20al.%22%2C%22parsedDate%22%3A%222020%22%2C%22numChildren%22%3A1%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20padding-left%3A%201em%3B%20text-indent%3A-1em%3B%26quot%3B%26gt%3B%5Cn%20%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%26gt%3BHalawa%2C%20M.%2C%20W%26%23xF6%3Bllhaf%2C%20M.%2C%20Vellasques%2C%20E.%2C%20Sanz%2C%20U.%20S.%2C%20%26amp%3B%20Hellwich%2C%20O.%20%282020%29.%20%26lt%3Bi%26gt%3BLearning%20Disentangled%20Expression%20Representations%20from%20Facial%20Images%26lt%3B%5C%2Fi%26gt%3B.%20WiCV%20at%20ECCV2020.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-DOIURL%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2008.07001%26%23039%3B%26gt%3Bhttps%3A%5C%2F%5C%2Fdoi.org%5C%2F10.48550%5C%2FarXiv.2008.07001%26lt%3B%5C%2Fa%26gt%3B%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22presentation%22%2C%22title%22%3A%22Learning%20Disentangled%20Expression%20Representations%20from%20Facial%20Images%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22presenter%22%2C%22firstName%22%3A%22Marah%22%2C%22lastName%22%3A%22Halawa%22%7D%2C%7B%22creatorType%22%3A%22presenter%22%2C%22firstName%22%3A%22Manuel%22%2C%22lastName%22%3A%22W%5Cu00f6llhaf%22%7D%2C%7B%22creatorType%22%3A%22presenter%22%2C%22firstName%22%3A%22Eduardo%22%2C%22lastName%22%3A%22Vellasques%22%7D%2C%7B%22creatorType%22%3A%22presenter%22%2C%22firstName%22%3A%22Urko%20S%5Cu00e1nchez%22%2C%22lastName%22%3A%22Sanz%22%7D%2C%7B%22creatorType%22%3A%22presenter%22%2C%22firstName%22%3A%22Olaf%22%2C%22lastName%22%3A%22Hellwich%22%7D%5D%2C%22abstractNote%22%3A%22Face%20images%20are%20subject%20to%20many%20different%20factors%20of%20variation%2C%20especially%20in%20unconstrained%20in-the-wild%20scenarios.%20For%20most%20tasks%20involving%20such%20images%2C%20e.g.%20expression%20recognition%20from%20video%20streams%2C%20having%20enough%20labeled%20data%20is%20prohibitively%20expensive.%20One%20common%20strategy%20to%20tackle%20such%20a%20problem%20is%20to%20learn%20disentangled%20representations%20for%20the%20different%20factors%20of%20variation%20of%20the%20observed%20data%20using%20adversarial%20learning.%20In%20this%20paper%2C%20we%20use%20a%20formulation%20of%20the%20adversarial%20loss%20to%20learn%20disentangled%20representations%20for%20face%20images.%20The%20used%20model%20facilitates%20learning%20on%20single-task%20datasets%20and%20improves%20the%20state-of-the-art%20in%20expression%20recognition%20with%20an%20accuracy%20of60.53%25on%20the%20AffectNetdataset%2C%20without%20using%20any%20additional%20data.%22%2C%22date%22%3A%222020%22%2C%22sessionTitle%22%3A%22%22%2C%22DOI%22%3A%2210.48550%5C%2FarXiv.2008.07001%22%2C%22citationKey%22%3A%22%22%2C%22url%22%3A%22%22%2C%22language%22%3A%22%22%2C%22collections%22%3A%5B%5D%2C%22dateModified%22%3A%222026-09-11T07%3A21%3A46Z%22%7D%7D%5D%7D
Halawa, M., Blume, F., Bideau, P., Maier, M., Rahman, R. A., & Hellwich, O. (2024). Multi-Task Multi-Modal Self-Supervised Learning for Facial Expression Recognition. 2024 IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), 4604–4614. https://doi.org/10.1109/CVPRW63382.2024.00463
Halawa, M., Hellwich, O., & Bideau, P. (2022). Action-Based Contrastive Learning for Trajectory Prediction. In S. Avidan, G. Brostow, M. Cissé, G. M. Farinella, & T. Hassner (Eds.), Computer Vision – ECCV 2022 (Vol. 13699, pp. 143–159). Springer Nature Switzerland. https://doi.org/10.1007/978-3-031-19842-7_9
Halawa, M., Wöllhaf, M., Vellasques, E., Sanz, U. S., & Hellwich, O. (2020). Learning Disentangled Expression Representations from Facial Images. WiCV at ECCV2020. https://doi.org/10.48550/arXiv.2008.07001
For the Public -
Girls’ Day 2022
