AI4Media workshop on “New learning paradigms & distributed AI”

Last Tuesday, May 4th, 2021, AI4Media organised the first of a series of workshops, that aimed at sharing the work and progress of the project in different research activities.

The first workshop was focused on “New learning paradigms & distributed AI”, and its main objective was to address the current limitations of learning approaches and improve speed and performance, by looking not only into the new approaches going beyond the current achievements in deep learning, such as lifelong and continuous learning, moving beyond Variational Autoencoders (VAEs), manifold and transfer learning, neural architecture search, quantum-assisted learning, the fusion of evolutionary algorithms and deep learning, but also, the present solutions for decentralised and distributed computation.

The agenda included presentations from the AI4Media partners and two guest speakers:

  • Dr. Xavier Alameda-Pinedia from INRIA, which addresses the research direction of audio-visual fusion for human behaviour understanding.
  • Prof. Rita Cucchiara from University of Modena and Reggio Emilia, who will tackle the hot subject of sustainable and privacy-preserving action understanding in videos.

Recording of the workshop:

Agenda of the Workshop:

Why is AI4Media important at technological, societal, political levels?

The AI4Media project is expected to transform the European AI landscape with regard to media & society and strengthen Europe’s position on the global AI stage.

Strengthening Europe’s leadership in AI research

AI4Media reinforces Europe’s research capacity in AI by performing cutting-edge research that caters to the needs of the media sector, tested through seven industrial use cases that range from combating disinformation in social media and supporting journalists for news story creation, to game design and artistic co-creation.

Delivering innovations for the News Media Industry

Ai4Media aims to provide new AI-based solutions for the News media sector by developing novel tools for journalistic research, journalistic creation, news production and service packaging, bias and disinformation detection, and audience/content research and performance optimisation.

Safeguarding democracy and political discourse

Media manipulation and disinformation campaigns relying on manipulated content are increasingly used as instruments for influencing key democratic processes, undermining democratic institutions, and re-shaping public opinion. AI4Media addresses these problems by developing AI technology for detecting disinformation in social media, with a view to supporting journalistic fact-checking and verification workflows in news organisations.

Putting AI in the service of the society

The research community carries a particular responsibility to help protect European societies from the potential negative impact of AI. By bringing together a diverse community of scientific experts on AI, media organisations, and social scientists, AI4Media aims to promote a unique brand of Ethical AI, powered by European actors and promoted by European media industries.

AI4Media at the workshop “Towards a Global Taxonomy of Interpretable AI”

“Towards a Global Taxonomy of Interpretable AI” was a half-day workshop, supported by AI4Media, that was held on April 29th, where around 20 experts actively participated in the two sessions of the workshop: a round table discussion with short presentations and a panel session.

Invited talks were given by the experts to illustrate their perspective on Interpretable AI from the cognitive, social, ethical, and legal perspectives.

During the panel session, experts answered questions on interpretability toolboxes, ethical and philosophical concerns, and how these tools can be used in practice, for example in medicine.

The slides and streaming of the event are available on the event’s website HERE

AI4Media at the workshop “Building Interpretable AI for Digital Pathology”

A half-day workshop on “Building Interpretable AI for Digital Pathology” was held online for the Applied Machine Learning Days of Lausanne on April 27th.

The workshop, supported by AI4Media, offered an introduction to digital pathology presented by Prof. Inti Zloebec, and an overview of interpretability techniques for machine learning algorithms applied to digital pathology given by Mara Graziani.

Two hands-on sessions were hosted on how to apply interpretability techniques to histopathology data, basic and the latest approaches in the research field such as regression concept vectors and graph-based modeling.

More information about the event HERE

AI4Media’s presence at “European Vision for AI 2021”

European scientists, leaders, high-level experts, and the general public discussed the present and future of artificial intelligence in Europe.

On Thursday 22 April 2021, an important discussion on the present and future development of Artificial Intelligence (AI) in Europe, which involved key European scientists, leaders, high-level experts, and the general public, took place during the “European Vision for AI 2021” event, just a day after the European Commission published its European Approach to Artificial Intelligence. The event, organised by the VISION project consortium partners in cooperation with four networks of centres of excellence on AI (AI4Media, ELISE, TAILOR, HumanE-AI-Net), was motivated by the need to discuss with the general public how the European scientific community is currently planning to move European AI forward, to future success in a competitive environment increasingly dominated by the US and China.

The general public was able to follow the event online and was able to interact via chat and a number of polls. Participants also had the opportunity to choose from three parallel sessions focussing on society, industry and skills & training, depending on their particular interests. A diverse set of panelists, covering a broad range of stakeholders in AI, discussed how the new EU plans affect these areas of high public interest.

At the event, European citizens with an interest in AI had the chance to gain an overview of Europe’s position in the field of AI and to learn about its impact on Europe’s economy and society. Thus, “European Vision for AI 2021” represented the first opportunity for a public discussion on the world’s first legal framework on AI and the new Coordinated Plan with the Member States, announced by the European Commission the day before, following up on its vision for a European eco-system of excellence and trust in AI.

The round-table discussion provided a chance to react to the proposed regulation and action plan with Dita Charanzová, Vice-President of the European Parliament; Kristian Kersting, TU Darmstadt, Germany; Ieva Martinkenaite, VP AI, Telenor, Norway; and Gabriele Mazzini, DG CNECT, European Commission.

According to Ms. Charanzová the rules “are a good and positive step. We are now all digesting the proposal [..]. I welcome it very much but as I always say ‘the devil is in detail’.  So we have to look at every detail of these regulations”. Ms. Charanzová pointed out that there already is legislation in place so the new regulation is about filling the gaps.

The need to coordinate efforts between research, industry, and society was also stressed by Ieva Martinkenaite: “We need to create best practices, positive examples for industry, involving policymakers, involving academia so that we learn how to apply these practices of trustworthy AI in business. It is not about creating compliance and regulations and directives. It’s about helping us to implement it”.

Kristian Kersting also stressed the importance of maintaining excellence in research through coordinated actions and the creation of large-scale infrastructure, through the creation of a “CERN for AI”.

AI4Media’s video was shown during the event and the project also participated in the parallel session dedicated to “Skills & Training: Education and personal development in AI” where AI4Media’s partner, Prof. Ioannis Pitas from Aristotle University of Thessaloniki, was one of the speakers.

The recorded video of the “Skills & Training” session is available below:

All videos from the event are available at VISION’s website HERE

AI4Media endorsed the Code-against-hate hackathon 2021

AI4Media endorsed and supported the Code-against-hate hackathon that was carried out on March 19-21, 2021. In particular, AI4Media’s partner, the AI Multimedia Lab of University Politehnica of Bucharest participated with a team in the hackathon and Prof. Bogdan Ionescu was a Mentor in the event.

Scope of the event

Public communicators, influencers, popular brands, journalists, etc. are targeted with online hate speech on a daily basis, particularly on Facebook and Twitter, when creating content. The amount of user-generated comments on these platforms is huge and identifying hate speech manually has become such a time-consuming task, that most public communicators feel discouraged to tackle it.

Let´s empower public communicators and social media users in their infinite fight against hate speech. The real challenge is not only to detect hate speech effectively but to develop a solution that would make moderation of online debate containing hate speech easier.

More information about the event HERE

AI4Media on Predicting Media Memorability at MediaEval 2021

On 11, 14 and 15 of December 2020, AI4Media supported the organisation of a workshop on “Predicting Media Memorability” at MediaEval 2021.

Media platforms such as social networks, media advertisement, information retrieval and recommendation systems deal with exponentially growing data day after day. Enhancing the relevance of multimedia occurrences in our everyday life requires new ways to organize – in particular, to retrieve – digital content. Like other metrics of video importance, such as aesthetics or interestingness, memorability can be regarded as useful to help make a choice between competing videos. This is even truer when one considers the specific use cases of creating commercials or creating educational content. Because the impact of different multimedia content, images or videos, on human memory is unequal, the capability of predicting the memorability level of a given piece of content is obviously of high importance for professionals in the field of advertising. Beyond advertising, other applications, such as filmmaking, education, content retrieval, etc., may also be impacted by the proposed task.

The task addressed in this workshop requires participants to automatically predict memorability scores for videos, that reflect the probability for a video to be remembered. Participants received an extensive data set of videos with memorability annotations, related information, and pre-extracted state-of-the-art visual features.

The AI4Media’s partners involved in this workshop were Mihai Gabriel Constantin and Bogdan Ionescu from Multimedia Lab, UPB and Claire-Hélène Demarty, InterDigital.

More information about the event HERE

AI4Media Workshop on GANs for Media Content Generation

On October 1st, 2020, AI4Media organised an online workshop on “GANs for Media Content Generation”, with the objective of looking into the use of Generative Adversarial Networks (GANs) for Media production and related challenges.

The workshop was moderated by AI4Media’s Coordinator Yiannis Kompatsiaris, and the following topics were addressed by the speakers from AI4Media’s consortium.

  • Learning to Predict Pixels Using AI for Content Enhancement and Delivery
  • Deepfake Detection: The Importance of Training Data Pre-processing and Practical Considerations
  • Image and Video Generation: A deep Learning Approach
  • Adversarial Face De-identification for Privacy Protection
  • Major Challenges in the Detection of Synthetic Media and Deepfakes

Watch the recording of this workshop below:

Access to all the presentations HERE

COOKIES

AI4Media may use cookies to store your login data, collect statistics to optimize the website's functionality and to perform marketing actions based on your interests.

COOKIES
They allow you to browse the website and use its applications as well as to access secure areas of the website. Without these cookies, the services you have requested cannot be provided.
These cookies are necessary to allow the main functionality of the website and they are activated automatically when you enter this website. They store user preferences for site usage so that you do not need to reconfigure the site each time you visit it.
These cookies direct advertising according to the interests of each user so as to direct advertising campaigns, taking into account the tastes of users, and they also limit the number of times you see the ad, helping to measure the effectiveness of advertising and the success of the website organisation.

Required Cookies They allow you to browse the website and use its applications as well as to access secure areas of the website. Without these cookies, the services you have requested cannot be provided.

Functional Cookies These cookies are necessary to allow the main functionality of the website and they are activated automatically when you enter this website. They store user preferences for site usage so that you do not need to reconfigure the site each time you visit it.

Advertising Cookies These cookies direct advertising according to the interests of each user so as to direct advertising campaigns, taking into account the tastes of users, and they also limit the number of times you see the ad, helping to measure the effectiveness of advertising and the success of the website organisation.