| | Welcome to the final ROXANNE newsletter | | A Goodbye Message from the Project Coordinator
Dear Readers,
The ROXANNE research project (started in September 2019) was envisaged with the objective of providing law enforcement agencies (LEAs) with a platform integrating advanced speech, text, video, and network analysis technologies that could be used lawfully to uncover criminal networks while minimizing investigation time and effort. The technical development was centered around an innovative platform, with the intention not to replace humans but to automate time-consuming tasks and support LEA decision-making.
When the project was launched at the kick-off meeting in 2019 at IDIAP in Switzerland, the consortium had a large number of challenges ahead (both legal and technical). These included unknown challenges such as data availability (given limited access to real-world data), and the need to adapt to pandemic constraints; organizing the first field test in a 100% remote format and switching to online dissemination for a large part of the project lifespan. Now that most of the planned project outcomes have become tangible results, it is a great pleasure to see how the project progressed despite these obstacles. We have achieved a lot as a multidisciplinary consortium of 25 partners from 15 countries, bringing together 11 LEAs, as well as representatives from industry and academia. Together we explored the operational, technical, legal, societal and ethical perspectives related to the evolving field of advanced data analysis technologies. In particular, we worked hard to ensure that the final solution meets the operational needs of law enforcement. We have successfully conducted three field tests and organized the Final Conference. After three years of collaborative work, we can present the results. The main outcome is the Autocrime platform, which is a unique solution to LEA problems, and will be made available to interested European LEAs free of charge. The second outcome that we are particularly proud of is the ROXSD - a synthetic, but highly realistic, dataset of communication data in a fictional organized crime network. This will be made available to other researchers in the fighting crime and terrorism field (known as FCT). Thank you for being with us for these last three years, I hope you found news from us interesting and please get in touch if you have any questions related to the project. Yours faithfully, Dr Petr Motlicek Project Coordinator | | | | The Final Conference – a big success!
Hosted by Capgemini Technology Services at Campus Cyber, Paris and online, the ROXANNE Final Conference turned out to be a big success thanks to all project partners involved. We presented the project results to 109 people, who joined us in person and online. Participants came from all over the world, including North and South America, Europe and Asia, with many different backgrounds represented – we were happy to welcome the ROXANNE consortium partners, LEAs, researchers, policy makers, representatives from industry, civil society organizations and our External Ethics Board.
| | | | | ROXANNE Final Video
Our latest video has just been released! It summarises the project and focuses on results that we have obtained during the project. You can watch the video here. | | | | | |
Blogs: - Training Speaker Embedding Extractors Using Multi-Speaker Audio with Unknown Speaker Boundaries (Themos Stafylakis, Ladislav Mosner, Oldrich Plchot, Johan Rohdin, Anna Silnova, Lukas Burget, Jan “Honza” Cernocky, (Interspeech 2022))
- A Few Thousand Translations Go a Long Way! Leveraging Pre-trained Models for African News Translation (David Adelani, Jesujoba Alabi, Angela Fan, Julia Kreutzer, Xiaoyu Shen, Machel Reid, Dana Ruiter, Dietrich Klakow, Peter Nabende, Ernie Chang, Tajuddeen Gwadabe, Freshia Sackey, Bonaventure F. P. Dossou, Chris Emezue, Colin Leong, Michael Beukman, Shamsuddeen Muhammad, Guyo Jarso, Oreen Yousuf, Andre Niyongabo Rubungo, Gilles Hacheme, Eric Peter Wairagala, Muhammad Umair Nasir, Benjamin Ajibade, Tunde Ajayi, Yvonne Gitau, Jade Abbott, Mohamed Ahmed, Millicent Ochieng, Anuoluwapo Aremu, Perez Ogayo, Jonathan Mukiibi, Fatoumata Ouoba Kabore, Godson Kalipe, Derguene Mbaye, Allahsera Auguste Tapo, Victoire Memdjokam Koagne, Edwin Munkoh-Buabeng, Valencia Wagner, Idris Abdulmumin, Ayodele Awokoya, Happy Buzaaba, Blessing Sibanda, Andiswa Bukula, Sam Manthalu, (NAACL-2022))
- TOKEN is a MASK: Few-shot Named Entity Recognition with Pre-trained Language Models (Ali Davody, David Ifeoluwa Adelani, Thomas Kleinbauer, Dietrich Klakow (25th International Conference on Text, Speech and Dialogue (TSD 2022))
- Task-Adaptive Pre-Training for Boosting Learning With Noisy Labels: A Study on Text Classification for African Languages (Dawei Zhu, Michael A. Hedderich, Fangzhou Zhai, David Ifeoluwa Adelani, Dietrich Klakow (AfricaNLP@ICLR 2022))
- Is BERT Robust to Label Noise? A Study on Learning with Noisy Labels in Text Classification (Dawei Zhu, Michael A. Hedderich, Fangzhou Zhai, David Ifeoluwa Adelani & Dietrich Klakow (Insights @ACL 2022))
| | | | | This project has received funding from the European Union’s Horizon 2020 Work Programme for research and innovation 2018-2020, under grant agreement n°833635. | | | | | | | |