Machine Learning for Multimedia Material.

Many approaches are used to extract relevant information from multimedia material, ranging from very low-level to higher-level descriptions (classes, captions, ...). That diversity of information is produced by algorithms that have varying degrees of supervision. Lately, fully supervised approaches based on deep learning proved to outperform most older techniques. This is particularly true for the latest developments of Recurrent Neural Networkds (RNN, such as LSTMs) or convolutional neural network (CNNs) for images that reach excellent performance  65. Linkmedia contributes to advancing the state of the art in computing representations for multimedia material by investigating the topics listed below. Some of them go beyond the very processing of multimedia material as they also question the fundamentals of machine learning procedures when applied to multimedia.

• Learning from few samples/weak supervisions. CNNs and RNNs need large collections of carefully annotated data. They are not fitted for analyzing datasets where few examples per category are available or only cheap image-level labels are provided. Linkmedia investigates low-shot, semi-supervised and weakly supervised learning processes: Augmenting scarce training data by automatically propagating labels  68, or transferring what was learned on few very well annotated samples to allow the precise processing of poorly annotated data  77. Note that this context also applies to the processing of heritage collections (paintings, illuminated manuscripts, ...) that strongly differ from contemporary natural images. Not only annotations are scarce, but the learning processes must cope with material departing from what standard CNNs deal with, as classes such as "planes", "cars", etc, are irrelevant in this case.
• Ubiquitous Training. NN (CNNs, LSTMs) are mainstream for producing representations suited for high-quality classification. Their training phase is ubiquitous because the same representations can be used for tasks that go beyond classification, such as retrieval, few-shot, meta- and incremental learning, all boiling down to some form of metric learning. We demonstrated that this ubiquitous training is relatively simpler  68 yet as powerful as ad-hoc strategies fitting specific tasks  81. We study the properties and the limitations of this ubiquitous training by casting metric learning as a classification problem.
• Beyond static learning. Multimedia collections are by nature continuously growing, and ML processes must adapt. It is not conceivable to re-train a full new model at every change, but rather to support continuous training and/or allowing categories to evolve as the time goes by. New classes may be defined from only very few samples, which links this need for dynamicity to the low-shot learning problem discussed here. Furthermore, active learning strategies determining which is the next sample to use to best improve classification must be considered to alleviate the annotation cost and the re-training process  72. Eventually, the learning process may need to manage an extremely large number of classes, up to millions. In this case, there is a unique opportunity of blending the expertise of Linkmedia on large scale indexing and retrieval with deep learning. Base classes can either be "summarized" e.g. as a multi-modal distribution, or their entire training set can be made accessible as an external associative memory  87.
• Learning and lightweight architectures. Multimedia is everywhere, it can be captured and processed on the mobile devices of users. It is necessary to study the design of lightweight ML architectures for mobile and embedded vision applications. Inspired by  91, we study the savings from quantizing hyper-parameters, pruning connections or other approximations, observing the trade-off between the footprint of the learning and the quality of the inference. Once strategy of choice is progressive learning which early aborts when confident enough  73.
• Multimodal embeddings. We pursue pioneering work of Linkmedia on multimodal embedding, i.e., representing multiple modalities or information sources in a single embedded space  85, 84, 86. Two main directions are explored: exploiting adversarial architectures (GANs) for embedding via translation from one modality to another, extending initial work in  86 to highly heterogeneous content; combining and constraining word and RDF graph embeddings to facilitate entity linking and explanation of lexical co-occurrences  62.
• Accountability of ML processes. ML processes achieve excellent results but it is mandatory to verify that accuracy results from having determined an adequate problem representation, and not from being abused by artifacts in the data. Linkmedia designs procedures for at least explaining and possibly interpreting and understanding what the models have learned. We consider heat-maps materializing which input (pixels, words) have the most importance in the decisions  80, Taylor decompositions to observe the individual contributions of each relevance scores or estimating LID 49 as a surrogate for accounting for the smoothness of the space.
• Extracting information. ML is good at extracting features from multimedia material, facilitating subsequent classification, indexing, or mining procedures. Linkmedia designs extraction processes for identifying parts in the images  78, 79, relationships between the various objects that are represented in images  55, learning to localizing objects in images with only weak, image-level supervision  80 or fine-grained semantic information in texts  60. One technique of choice is to rely on generative adversarial networks (GAN) for learning low-level representations. These representations can e.g. be based on the analysis of density  90, shading, albedo, depth, etc.
• Learning representations for time evolving multimedia material. Video and audio are time evolving material, and processing them requests to take their time line into account. In  74, 58 we demonstrated how shapelets can be used to transform time series into time-free high-dimensional vectors, preserving however similarities between time series. Representing time series in a metric space improves clustering, retrieval, indexing, metric learning, semi-supervised learning and many other machine learning related tasks. Research directions include adding localization information to the shapelets, fine-tuning them to best fit the task in which they are used as well as designing hierarchical representations.

Adversarial Machine Learning.

Systems based on ML take more and more decisions on our behalf, and maliciously influencing these decisions by crafting adversarial multimedia material is a potential source of dangers: a small amount of carefully crafted noise imperceptibly added to images corrupts classification and/or recognition. This can naturally impact the insight users get on the multimedia collection they work with, leading to taking erroneous decisions e.g.

This adversarial phenomenon is not particular to deep learning, and can be observed even when using other ML approaches  54. Furthermore, it has been demonstrated that adversarial samples generalize very well across classifiers, architectures, training sets. The reasons explaining why such tiny content modifications succeed in producing severe errors are still not well understood.

We are left with little choice: we must gain a better understanding of the weaknesses of ML processes, and in particular of deep learning. We must understand why attacks are possible as well as discover mechanisms protecting ML against adversarial attacks (with a special emphasis on convolutional neural networks). Some initial contributions have started exploring such research directions, mainly focusing on images and computer vision problems. Very little has been done for understanding adversarial ML from a multimedia perspective  59.

Linkmedia is in a unique position to throw at this problem new perspectives, by experimenting with other modalities, used in isolation one another, as well as experimenting with true multimodal inputs. This is very challenging, and far more complicated and interesting than just observing adversarial ML from a computer vision perspective. No one clearly knows what is at stake with adversarial audio samples, adversarial video sequences, adversarial ASR, adversarial NLP, adversarial OCR, all this being often part of a sophisticated multimedia processing pipeline.

Our ambition is to lead the way for initiating investigations where the full diversity of modalities we are used to work with in multimedia are considered from a perspective of adversarial attacks and defenses, both at learning and test time. In addition to what is described above, and in order to trust the multimedia material we analyze and/or the algorithms that are at play, Linkmedia investigates the following topics:

• Beyond classification. Most contributions in relation with adversarial ML focus on classification tasks. We started investigating the impact of adversarial techniques on more diverse tasks such as retrieval  48. This problem is related to the very nature of euclidean spaces where distances and neighborhoods can all be altered. Designing defensive mechanisms is a natural companion work.
• Detecting false information. We carry-on with earlier pioneering work of Linkmedia on false information detection in social media. Unlike traditional approaches in image forensics  63, we build on our expertise in content-based information retrieval to take advantage of the contextual information available in databases or on the web to identify out-of-context use of text or images which contributed to creating a false information  75.
• Deep fakes. Progress in deep ML and GANs allow systems to generate realistic images and are able to craft audio and video of existing people saying or doing things they never said or did 71. Gaining in sophistication, these machine learning-based "deep fakes" will eventually be almost indistinguishable from real documents, making their detection/rebutting very hard. Linkmedia develops deep learning based counter-measures to identify such modern forgeries. We also carry on with making use of external data in a provenance filtering perspective  92 in order to debunk such deep fakes.
• Distributions, frontiers, smoothness, outliers. Many factors that can possibly explain the adversarial nature of some samples are in relation with their distribution in space which strongly differs from the distribution of natural, genuine, non adversarial samples. We are investigating the use of various information theoretical tools that facilitate observing distributions, how they differ, how far adversarial samples are from benign manifolds, how smooth is the feature space, etc. In addition, we are designing original adversarial attacks and develop detection and curating mechanisms  49.

Multimedia Knowledge Extraction.

Information obtained from collections via computer ran processes is not the only thing that needs to be represented. Humans are in the loop, and they gradually improve their level of understanding of the content and nature of the multimedia collection. Discovering knowledge and getting insight is involving multiple people across a long period of time, and what each understands, concludes and discovers must be recorded and made available to others. Collaboratively inspecting collections is crucial. Ontologies are an often preferred mechanism for modeling what is inside a collection, but this is probably limitative and narrow.

Linkmedia is concerned with making use of existing strategies in relation with ontologies and knowledge bases. In addition, Linkmedia uses mechanisms allowing to materialize the knowledge gradually acquired by humans and that might be subsequently used either by other humans or by computers in order to better and more precisely analyze collections. This line of work is instantiated at the core of the iCODA project Linkmedia coordinates.

We are therefore concerned with:

• Multimedia analysis and ontologies. We develop approaches for linking multimedia content to entities in ontologies for text and images, building on results in multimodal embedding to cast entity linking into a nearest neighbor search problem in a high-dimensional joint embedding of content and entities  84. We also investigate the use of ontological knowledge to facilitate information extraction from content  62.
• Explainability and accountability in information extraction. In relation with ontologies and entity linking, we develop innovative approaches to explain statistical relations found in data, in particular lexical or entity co-occurrences in textual data, for example using embeddings constrained with translation properties of RDF knowledge or path-based explanation within RDF graphs. We also work on confidence measures in entity linking and information extraction, studying how the notions of confidence and information source can be accounted for in knowledge basis and used in human-centric collaborative exploration of collections.
• Dynamic evolution of models for information extraction. In interactive exploration and information extraction, e.g., on cultural or educational material, knowledge progressively evolves as the process goes on, requiring on-the-fly design of new models for content-based information extractors from very few examples, as well as continuous adaptation of the models. Combining in a seamless way low-shot, active and incremental learning techniques is a key issue that we investigate to enable this dynamic mechanisms on selected applications.

Searching.

Search engines must run similarity searches very efficiently. High-dimensional indexing techniques therefore play a central role. Yet, recent contributions in ML suggest to revisit indexing in order to adapt to the specific properties of modern features describing contents.

• Advanced scalable indexing. High-dimensional indexing is one of the foundations of Linkmedia. Modern features extracted from the multimedia material with the most recent ML techniques shall be indexed as well. This, however, poses a series of difficulties due to the dimensionality of these features, their possible sparsity, the complex metrics in use, the task in which they are involved (instance search, $k$-nn, class prototype identification, manifold search  68, time series retrieval, ...). Furthermore, truly large datasets require involving sketching  52, secondary storage and/or distribution  51, 50, alleviating the explosion of the number of features to consider due to their local nature or other innovative methods  67, all introducing complexities. Last, indexing multimodal embedded spaces poses a new series of challenges.
• Improving quality. Scalable indexing techniques are approximate, and what they return typically includes a fair amount of false positives. Linkmedia works on improving the quality of the results returned by indexing techniques. Approaches taking into account neighborhoods  61, manifold structures instead of pure distance based similarities  68 must be extended to cope with advanced indexing in order to enhance quality. This includes feature selection based on intrinsic dimensionality estimation  49.
• Dynamic indexing. Feature collections grow, and it is not an option to fully reindex from scratch an updated collection. This trivially applies to the features directly extracted from the media items, but also to the base class prototypes that can evolve due to the non-static nature of learning processes. Linkmedia will continue investigating what is at stake when designing dynamic indexing strategies.

Navigating.

Navigating a multimedia collection is very central to its understanding. It differs from searching as navigation is not driven by any specific query. Rather, it is mostly driven by the relationships that various documents have one another. Relationships are supported by the links between documents and/or parts of documents. Links rely on semantic similarity, depicting the fact that two documents share information on the same topic. But other aspects than semantics are also at stake, e.g., time with the dates of creation of the documents or geography with mentions or appearance in documents of some geographical landmarks or with geo-tagged data.

In multimedia collections, links can be either implicit or explicit, the latter being much easier to use for navigation. An example of an implicit link can be the name of someone existing in several different news articles; we, as humans, create a mental link between them. In some cases, the computer misses such configurations, leaving such links implicit. Implicit links are subject to human interpretation, hence they are sometimes hard to identify for any automatic analysis process. Implicit links not being materialized, they can therefore hardly be used for navigation or faceted search. Explicit links can typically be seen as hyperlinks, established either by content providers or, more aligned with Linkmedia, automatically determined from content analysis. Entity linking (linking content to an entity referenced in a knowledge base) is a good example of the creation of explicit links. Semantic similarity links, as investigated in the LIMAH project and as considered in the search and hyperlinking task at MediaEval and TRECVid, are also prototypical links that can be made explicit for navigation. Pursuing work, we investigate two main issues:

• Improving multimodal content-based linking. We exploit achievements in entity linking to go beyond lexical or lexico-visual similarity and to provide semantic links that are easy to interpret for humans; carrying on, we work on link characterization, in search of mechanisms addressing link explainability (i.e., what is the nature of the link), for instance using attention models so as to focus on the common parts of two documents or using natural language generation; a final topic that we address is that of linking textual content to external data sources in the field of journalism, e.g., leveraging topic models and cue phrases along with a short description of the external sources.
• Dynamicity and user-adaptation. One difficulty for explicit link creation is that links are often suited for one particular usage but not for another, thus requiring creating new links for each intended use; whereas link creation cannot be done online because of its computational cost, the alternative is to generate (almost) all possible links and provide users with selection mechanisms enabling personalization and user-adaptation in the exploration process; we design such strategies and investigate their impact on exploration tasks in search of a good trade-off between performance (few high-quality links) and genericity.

Summarizing.

Multimedia collections contain far too much information to allow any easy comprehension. It is mandatory to have facilities to aggregate and summarize a large body on information into a compact, concise and meaningful representation facilitating getting insight. Current technology suggests that multimedia content aggregation and story-telling are two complementary ways to provide users with such higher-level views. Yet, very few studies already investigated these issues. Recently, video or image captioning  88, 83 have been seen as a way to summarize visual content, opening the door to state-of-the-art multi-document text summarization  64 with text as a pivot modality. Automatic story-telling has been addressed for highly specific types of content, namely TV series  56 and news  76, 82, but still need a leap forward to be mostly automated, e.g., using constraint-based approaches for summarization  53, 82.

Furthermore, not only the original multimedia material has to be summarized, but the knowledge acquired from its analysis is also to summarize. It is important to be able to produce high-level views of the relationships between documents, emphasizing some structural distinguishing qualities. Graphs establishing such relationships need to be constructed at various level of granularity, providing some support for summarizing structural traits.

Summarizing multimedia information poses several scientific challenges that are:

• Choosing the most relevant multimedia aggregation type: Taking a multimedia collection into account, a same piece of information can be present in several modalities. The issue of selecting the most suitable one to express a given concept has thus to be considered together with the way to mix the various modalities into an acceptable production. Standard summarization algorithms have to be revisited so that they can handle continuous representation spaces, allowing them to benefit from the various modalities  57.
• Expressing user’s preferences: Different users may appreciate quite different forms of multimedia summaries, and convenient ways to express their preferences have to be proposed. We for example focus on the opportunities offered by the constraint-based framework.
• Evaluating multimedia summaries: Finding criteria to characterize what a good summary is remains challenging, e.g., how to measure the global relevance of a multimodal summary and how to compare information between and across two modalities. We tackle this issue particularly via a collaboration with A. Smeaton at DCU, comparing the automatic measures we will develop to human judgments obtained by crowd-sourcing;
• Taking into account structuring and dynamicity: Typed links between multimedia fragments, and hierarchical topical structures of documents obtained via work previously developed within the team are two types of knowledge which have seldom been considered as long as summarization is concerned. Knowing that the event present in a document is causally related to another event described in another document can however modify the ways summarization algorithms have to consider information. Moreover the question of producing coarse-to-fine grain summaries exploiting the topical structure of documents is still an open issue. Summarizing dynamic collections is also challenging and it is one of the questions we consider.

Mobile search

As part of our involvement in innovation project MobilAI, we have developed a novel knowledge transfer mechanism for metric learning 45, which can train a lightweight student network for image retrieval in a teacher-student setting, allowing it to outperform a large teacher network.

Our work is truly motivated by working together with a number of startup companies on mobile visual recognition. The companies have well-established technologies involving visual search, including for instance copyright protection by watermarking, worldwide identity document recognition and augmented reality in exhibitions.

However, solutions are mostly off-line or web-based; when mobile, they are mostly based on shallow representations, which still perform better than very small deep networks. Mobile and embedded computer vision applications are expected to have significant impact especially in developing countries, where access to computing is limited otherwise.

Despite the progress in efficient architectures, making small networks perform as well as large ones in different tasks is an enabling factor for mobile computing that is under-explored. While striving for scientific novelty, the interest of startup companies in our work for the development of innovative solutions is a direct indicator of socioeconomic impact to us.

7.1.1 TagEx

• Name: Yet another Part-of-Speech Tagger for French
• Keyword: Natural language processing
• Functional Description: TagEx is available as a web-service on https://allgo.inria.fr . Refer to Allgo for its usage.
• URL: https://allgo.inria.fr/app/tagex
• Contact: Vincent Claveau

7.1.2 NegDetect

• Name: Negation Detection
• Keyword: Natural language processing
• Functional Description: NegDetect relies on several layers of machine learning techniques (CRF, neural networks).
• Contacts: Vincent Claveau, Clément Dalloux

7.1.3 SurFree

• Name: A fast surrogate-free black-box attack against classifier
• Keywords: Computer vision, Classification, Cyber attack
• Scientific Description:

  Machine learning classifiers are critically prone to evasion attacks. Adversarial examples are slightly modified inputs that are then misclassified, while remaining perceptively close to their originals. Last couple of years have witnessed a striking decrease in the amount of queries a black box attack submits to the target classifier, in order to forge adversarials. This particularly concerns the blackbox score-based setup, where the attacker has access to top predicted probabilites: the amount of queries went from to millions of to less than a thousand.

  This paper presents SurFree, a geometrical approach that achieves a similar drastic reduction in the amount of queries in the hardest setup: black box decision-based attacks (only the top-1 label is available). We first highlight that the most recent attacks in that setup, HSJA, QEBA and GeoDA all perform costly gradient surrogate estimations. SurFree proposes to bypass these, by instead focusing on careful trials along diverse directions, guided by precise indications of geometrical properties of the classifier decision boundaries. We motivate this geometric approach before performing a head-to-head comparison with previous attacks with the amount of queries as a first class citizen. We exhibit a faster distortion decay under low query amounts (few hundreds to a thousand), while remaining competitive at higher query budgets.

  Paper : https://arxiv.org/abs/2011.12807

• Functional Description: This software is the implementation in python of the attack SurFree. This is an attack against a black-box classifier. It finds an input close to the reference input (Euclidean distance) yet not classified with the same predicted label as the reference input. This attack has been tested against image classifier in computer vision.
• URL: https://github.com/t-maho/SurFree
• Authors: Thibault Maho, Erwan Le Merrer, Teddy Furon
• Contacts: Teddy Furon, Thibault Maho, Erwan Le Merrer

7.1.4 GrowAndPrune

• Name: Neural architecture growing, pruning and search
• Keywords: Deep learning, Neural architecture search
• Functional Description: This is the official code that enables the reproduction of the results of our work https://avrithis.net/data/cv/pdf/msc/2020.neitthoffer.pdf
• URL: https://github.com/shymine/neural-architecture-growing-pruning-and-search
• Contacts: Timothee Neitthoffer, Ioannis Avrithis

7.1.5 AML

• Name: Asymmetric Metric Learning
• Keywords: Knowledge transfer, Metric learning, Image retrieval
• Functional Description: This is the official code and a set of pre-trained models that enable the reproduction of the results of our paper https://hal.inria.fr/hal-03047591.
• URL: https://github.com/budnikm/asymmetric_metric_learning
• Contacts: Mateusz Budnik, Ioannis Avrithis

7.1.6 NFSL

• Name: Noisy Few-Shot Learning
• Keywords: Few-shot learning, Deep learning
• Functional Description: This is the official code that enables the reproduction of the results of our paper https://hal.inria.fr/hal-03047513.
• URL: https://github.com/google-research/noisy-fewshot-learning
• Contacts: Ahmet Iscen, Ioannis Avrithis

7.1.7 DSM

• Name: Deep Spatial Matching
• Keywords: Spatial matching, Content-based Image Retrieval, Deep learning
• Functional Description: This is the official code that enables the reproduction of the results of our paper https://hal.inria.fr/hal-02374156.
• URL: https://github.com/osimeoni/DSM
• Contacts: Oriane Simeoni, Ioannis Avrithis

7.1.8 DAL

• Name: Rethinking Deep Active Learning
• Keywords: Active Learning, Deep learning
• Functional Description: This is the official code that enables the reproduction of the results of our paper https://hal.inria.fr/hal-02372102.
• URL: https://github.com/osimeoni/RethinkingDeepActiveLearning
• Contacts: Oriane Simeoni, Mateusz Budnik, Ioannis Avrithis

8.1.1 Building Medical concept embeddings without texts

Participants: Vincent Claveau.

8.1.2 CAS: corpus of clinical cases in French

Participants: Clément Dalloux, Vincent Claveau, Natalia Grabar.

8.1.3 On the Correlation of Word Embedding Evaluation Metrics

Participants: François Torregrossa, Vincent Claveau, Nihel Kooli, Guillaume Gravier, Robin Allesiardo.

8.1.4 HierarX: a tool for discovering hierarchies in hyperbolic spaces

Participants: François Torregrossa, Guillaume Gravier, Vincent Claveau, Nihel Kooli.

This work 36 introduces the HierarX tool which projects multiple datasources into hyperbolicmanifolds : Lorentz or Poincaré. From similarities between word pairs or continuous wordrepresentations in high dimensional spaces, HierarX is able to embed knowledge in hyperbolicgeometries with small dimensionality. Those shape information into continuous hierarchies.This work presents the HierarX workflow as well as its main use-cases.

8.1.5 Few-Shot Few-Shot Learning and the role of Spatial Attention

Participants: Yann Lifchitz, Yannis Avrithis, Sylvaine Picard.

Few-shot learning is often motivated by the ability of humans to learn new tasks from few examples. However, standard few-shot classification benchmarks assume that the representation is learned on a limited amount of base class data, ignoring the amount of prior knowledge that a human may have accumulated before learning new tasks. At the same time, even if a powerful representation is available, it may happen in some domain that base class data are limited or non-existent. This motivates us to study a problem where the representation is obtained from a classifier pre-trained on a large-scale dataset of a different domain, assuming no access to its training process, while the base class data are limited to few examples per class and their role is to adapt the representation to the domain at hand rather than learn from scratch. We adapt the representation in two stages, namely on the few base class data if available and on the even fewer data of new tasks. In doing so, we obtain from the pre-trained classifier a spatial attention map that allows focusing on objects and suppressing background clutter. This is important in the new problem, because when base class data are few, the network cannot learn where to focus implicitly. We also show that a pre-trained network may be easily adapted to novel classes, without meta-learning 29.

8.1.6 Local Propagation for Few-Shot Learning

Participants: Yann Lifchitz, Yannis Avrithis, Sylvaine Picard.

8.1.7 Iterative label cleaning for transductive and semi-supervised few-shot learning

Participants: Michalis Lazarou, Yannis Avrithis, Tania Stathaki.

Focusing on these two settings, we introduce a new algorithm that leverages the manifold structure of the labeled and unlabeled data distribution to predict pseudo-labels, while balancing over classes and using the loss value distribution of a limited-capacity classifier to select the cleanest labels, iterately improving the quality of pseudo-labels 47. Our solution sets new state of the art on four benchmark datasets, namely miniImageNet, tieredImageNet, CUB and CIFAR-FS, while being robust over feature space pre-processing and the quantity of available data.

8.1.8 Graph Convolutional Networks for Learning with Few Clean and Many Noisy Labels

Participants: Ahmet Iscen, Giorgos Tolias, Yannis Avrithis, Ondra Chum, Cordelia Schmid.

8.1.9 Joint Learning of Assignment and Representation for Biometric Group Membership

Participants: Marzieh Gheisari Khorasgani, Teddy Furon, Laurent Amsaleg.

8.1.10 Interactive Learning for Multimedia at Large

Participants: Omar Shahbaz Khan, Björn Þór Jónsson, Stevan Rudinac, Jan Zahálka, Hanna Ragnarsdóttir, Þórhildur Þorleiksdóttir, Gylfi Þór Guðmundsson, Laurent Amsaleg, Marcel Worring.

8.1.11 Asymmetric Metric Learning for Knowledge Transfer

Participants: Mateusz Budnik, Yannis Avrithis.

8.1.12 Exploring Quality Camouflage for Social Images

Participants: Zhuoran Liu, Zhengyu Zhao, Martha Larson, Laurent Amsaleg.

8.1.13 Fooling an Automatic Image Quality Estimator

Participants: Benoît Bonnet, Teddy Furon, Patrick Bas.

8.1.14 High Intrinsic Dimensionality Facilitates Adversarial Attack: Theoretical Evidence

Participants: Laurent Amsaleg, James Bailey, Amélie Barbe, Sarah Erfani, Teddy Furon, Michael Houle, Miloš Radovanović, Vinh Nguyen Xuan.

Machine learning systems are vulnerable to adversarial attack. By applying to the input object a small, carefully-designed perturbation, a classifier can be tricked into making an incorrect prediction. This phenomenon has drawn wide interest, with many attempts made to explain it. However, a complete understanding is yet to emerge. In this work we adopt a slightly different perspective, still relevant to classification 8. We consider retrieval, where the output is a set of objects most similar to a user-supplied query object, corresponding to the set of k-nearest neighbors. We investigate the effect of adversarial perturbation on the ranking of objects with respect to a query. Through theoretical analysis, supported by experiments, we demonstrate that as the intrinsic dimensionality of the data domain rises, the amount of perturbation required to subvert neighborhood rankings diminishes, and the vulnerability to adversarial attack rises. We examine two modes of perturbation of the query: either 'closer' to the target point, or 'farther' from it. We also consider two perspectives: 'query-centric', examining the effect of perturbation on the query's own neighborhood ranking, and 'target-centric', considering the ranking of the query point in the target's neighborhood set. All four cases correspond to practical scenarios involving classification and retrieval.

8.1.15 An alternative proof of the vulnerability of k-NN classifiers in high intrinsic dimensionality regions

Participants: Teddy Furon.

8.1.16 Defending Adversarial Examples via DNN Bottleneck Reinforcement

Participants: Wenqing Liu, Miaojing Shi, Teddy Furon, Li Li.

8.1.17 What if Adversarial Samples were Digital Images?

Participants: Benoît Bonnet, Teddy Furon, Patrick Bas.

8.1.18 Smooth Adversarial Examples

Participants: Hanwei Zhang, Yannis Avrithis, Teddy Furon, Laurent Amsaleg.

8.1.19 Walking on the Edge: Fast, Low-Distortion Adversarial Examples

Participants: Hanwei Zhang, Yannis Avrithis, Teddy Furon, Laurent Amsaleg.

8.1.20 Adversarial Regularization for Explainable-by-Design Time Series Classification

Participants: Yichang Wang, Rémi Emonet, Elisa Fromont, Simon Malinowski, Romain Tavenard.

8.1.21 Detecting Human-Object Interaction with Mixed Supervision

Participants: Suresh Kumaraswamy, Miaojing Shi, Ewa Kijak.

8.1.22 A correlation-based entity embedding approach for robust entity linking

Participants: Cheikh Brahim El Vaigh, François Torregrossa, Robin Allesiardo, Guillaume Gravier, Pascale Sébillot.

Done as part of the IPL iCODA.

8.1.23 IRISA System for Entity Detection and Linking at CLEF HIPE 2020

Participants: Cheikh Brahim El Vaigh, Guillaume Le Noé-Bienvenu, Guillaume Gravier, Pascale Sébillot.

8.1.24 Relation, es-tu là ? Détection de relations par LSTM pour améliorer l’extraction de relations

Participants: Cyrielle Mallart, Michel Le Nouy, Guillaume Gravier, Pascale Sébillot.

8.1.25 Understanding the phenomenology of reading through modelling

Participants: Alessio Antonini, Mari Carmen Suárez-Figueroa, Alessandro Adamou, Francesca Benatti, François Vignale, Guillaume Gravier, Lucia Lupi.

8.1.26 Rethinking deep active learning: Using unlabeled data at model training

Participants: Oriane Siméoni, Mateusz Budnik, Yannis Avrithis, Guillaume Gravier.

8.1.27 Improving topic modeling through homophily for legal documents

Participants: Kazuki Ashihara, Cheikh Brahim El Vaigh, Chenhui Chu, Benjamin Renoust, Noriko Okubo, Noriko Takemura, Yuta Nakashima, Hajime Nagahara.

8.2.1 Detection of Fake News in Social Networks: the MediaEval2020 challenge

Participants: Vincent Claveau.

8.2.2 Supervised learning for the detection of negation and of its scope in French and Brazilian Portuguese biomedical corpora

Participants: Clément Dalloux, Vincent Claveau, Natalia Grabar, Lucas Emanuel Silva Oliveira, Claudia Maria Cabral Moro, Yohan Bonescki Gumiel, Deborah Ribeiro Carvalho.

8.2.3 Supervised Learning for the ICD-10 Coding of French Clinical Narratives

Participants: Clément Dalloux, Vincent Claveau, Marc Cuggia, Guillaume Bouzillé, Natalia Grabar.

8.2.4 Hierarchical Multi-Label Propagation using Speaking Face Graphs for Multimodal Person Discovery

Participants: Gabriel Barbosa da Fonseca, Gabriel Sargent, Ronan Sicre, Zenilton Kleber Gonçalves Do Patrocinio, Guillaume Gravier, Silvio Jamil Guimarães.

8.2.5 A Novel Path-based Entity Relatedness Measure for Efficient Collective Entity Linking

Participants: Cheikh Brahim El Vaigh, François Goasdoué, Guillaume Gravier, Pascale Sébillot.

CIFRE PhD: Incremental dynamic construction of knowledge bases from text mining

Participants: Guillaume Gravier, Cyrielle Mallart, Pascale Sébillot.

Duration: 3 years, started in Dec. 2018 Partner: Ouest France

In the context of a newspaper, the thesis explores the combination of text mining and knowledge representation techniques to assist the extraction, interpretation and validation of valuable pieces of information from the journal’s content so as to incrementally build a full-scale knowledge base. This thesis is in close relation with the iCODA Inria Project Lab, with direct contribution to the project’s results.

CIFRE PhD: Embedding heterogeneous data for directory search

Participants: Vincent Claveau, Guillaume Gravier, François Torregrossa.

Duration: 3 years, started in Dec. 2018 Partner: SoLocal

The thesis aims at learning how to jointly exploit heterogeneous sources of information (e.g., names, activity sector, user profiles, queries, etc.) in the design of neural network embeddings for information retrieval and language understanding. Applications cover natural language query analysis and personalized information retrieval in Pagesjaunes’ directory.

CIFRE PhD: Few shot learning for object recognition in aerial images

Participants: Yannis Avrithis, Yann Lifchitz.

Duration: 3 years, started in March 2018 Partner: Safran Tech

This is a CIFRE PhD thesis project aiming to study architectures and learning techniques most suitable for object recognition from few samples and to validate these approaches on multiple recognition tasks and use-cases related to aerial images.

CIFRE PhD: Deep Learning and Homomorphic encryption

Participants: Teddy Furon, Samuel Tap.

Duration: 3 years, started in December 2020 Partner: ZAMA.ia

This is a CIFRE PhD thesis project aiming to study inference and training of neural networks in the encrypted domain. This means that inputs (test or training data) are encrypted to protect confidentiality.

CIFRE PhD: Robustness of machine learning against uncertainties

Participants: Teddy Furon, Karim Tit.

Duration: 3 years, started in December 2020 Partner: THALES La Ruche

This is a CIFRE PhD thesis project aiming to study the robustness of machine learning algorithm facing uncertainties in the acquisition chain of the data.

CIFRE PhD: Semantic multimodal question answering in domestic environments

Participants: Teddy Furon, Deniz Engin.

Duration: 3 years, started in June 2020Partner: InterDigital

This is a CIFRE PhD thesis project aiming at designing novel deep learning based MQA techniques that takes in to account rich information from different sensors to facilitate living condition at home. Advances in artificial intelligence boost research towards VQA as well as multimodal analysis.

CIFRE PhD: Multimodal detection of fake news

Participants: Vincent Claveau, Ewa Kijak, Antoine Chaffin.

Duration: 3 years, started in November 2020 Partner: IMATAG

This is a CIFRE PhD thesis project aiming at designing multimodal models able to detect fake news, like repurposing techniques, based on joint analysis of visual and textual modalities.

CIFRE PhD: Semantic multimodal question answering (MQA) in domestic environments

Participants: Yannis Avrithis, Teddy Furon, Deniz Engin.

Duration: 3 years, started in September 2020 Partner: InterDigital

This is a CIFRE PhD thesis project aiming at designing novel question answering methods based on deep learning to facilitate living conditions in home environments. It investigates moving from image understanding towards multimodal context understanding in video of long duration. This may allow answering questions based on what has happened in the past.

10.1.1 Inria associate team not involved in an IIL

10.1.2 Inria international partners

10.1.3 Participation in other international programs

10.2.1 Visits of international scientists

Michalis Lazarou, PhD student at Imperial College, University of London. Planned to stay 5 months (November 2020 - January 2021), but left in November 2020 (stayed 2 months) due to health crisis.

Philip Bellos, MSc student at National and Kapodistrian University of Athens. Planned to stay 4 months (October 2020 - January 2021), but left in November 2020 (stayed 1 month) due to health crisis.

Vasileios Psomas, MSc student at National and Kapodistrian University of Athens. Stayed 4 months (February-May 2020).

Amaia Abanda, PhD Student at BCAM, Spain. She stayed from mid-September to end of October (3 months were planned). Josu Ircio Fernandez, PhD Student at the Center for Technological Research, Spain stayed in October (3 months were planned, but the stay was shortened due to health crisis).

10.3.1 Collaborations in European programs, except FP7 and H2020

Chaire Security of AI for Defense Applications (SAIDA)

Participants: Teddy Furon, Laurent Amsaleg, Erwan Le Merrer, Mathias Rousset, Benoit Bonnet, Thibault Maho, Patrick Bas, Samuel Tap, Karim Tit.

Duration: 4 years, started Sept 2020ANR-20-CHIA-0011-01

Inria Project Lab Knowledge-driven data and content collaborative analytics (iCODA)

Participants: Laurent Amsaleg, Cheikh Brahim El Vaigh, Guillaume Gravier, Cyrielle Mallart, Pascale Sébillot.

Duration: 4.5 years, started in April 2017 Partners: Inria project-teams Linkmedia, CEDAR, GraphIK and ILDA, with Ouest-France, Le Monde and AFP

One of today’s major issues in data science is the design of algorithms that allow analysts to efficiently infer useful information and knowledge by collaboratively inspecting heterogeneous information sources, from structured data to unstructured content. Taking data journalism as an emblematic use-case, the goal of the project is to develop the scientific and technological foundations for knowledge- mediated user-in-the-loop collaborative data analytics on heterogeneous information sources, and to demonstrate the effectiveness of the approach in realistic, high-visibility use-cases. The project stands at the crossroad of multiple research fields—content analysis, data management, knowledge representation, visualization—that span multiple Inria themes, and counts on a club of major press partners to define usage scenarios, provide data and demonstrate achievements.

INRIA-BNF: Classification d'images patrimoniales (CIP)

Participants: Florent Michel, Laurent Amsaleg, Guillaume Gravier, Ewa Kijak, Yannis Avrithis.

Duration: 1 year, started in Dec 2018. Extended to May 2020.

This project is within the context of the collaborations between INRIA and the French Ministry of Culture. In that context, we have started a collaboration with the French National Library (BNF) which collects, preserves and makes known the national documentary heritage. This collaboration aims at facilitating the automatic classification of heritage images through the use of recent deep-learning techniques. Such images are quite specific: they are not at all similar with what deep-learning techniques are used to work with, that is, the classification of heritage images does not target modern categories such as planes, cars, cats and dogs because this is irrelevant and because heritage collections do not include images of contemporary objects. Furthermore, heritage images come in vast quantities, but they are little annotated and deep-learning techniques can hardly rely on massive annotations to easily learn. Last, the learning has to be continuous as curators may need to add or modify existing classes, without re-learning everything from scratch.

The techniques of choice to reach that goal include the semi-supervised learning, low-shot learning techniques, knowledge transfer, fine tuning existing models, etc.

ANR Archival: Multimodal machine comprehension of language for new intelligent interfaces of scientific and cultural mediation

Participants: Laurent Amsaleg, Guillaume Gravier, Duc Hau Nguyen, Pascale Sébillot.

Duration: 3.5 year, started in Dec. 2019

The multidisciplinary and multi-actor ARCHIVAL project aims at yielding collaborations between researchers from the fields of Information and Communication Sciences as well as Computer Sciences around archive value enhancing and knowledge sharing for arts, culture and heritage. The project is structured around the following questionings: What part can machine comprehension methods play towards the reinterpretation of thematic archive collections? How can content mediation interfaces exploit results generated by current AI approaches?

ARCHIVAL teams will explore heterogeneous document collection structuration in order to explicitly reveal implicit links, to explain the nature of these links and to promote them in an intelligible way towards ergonomic mediation interfaces that will guarantee a successful appropriation of contents. A corpus has been delimited from the FMSH “self-management” collection, recently awarded as Collex, which will be completed from the large Canal-U academic audiovisual portal. The analysis and enhancement of this collection is of particular interest for Humanities and Social Sciences in a context where it becomes a necessity to structurally reconsider new models of socioeconomic development (democratic autonomy, social and solidarity-based economy, alternative development,…).

ANR MEERQAT: MultimEdia Entity Representation and Question Answering Tasks

Participants: Laurent Amsaleg, Yannis Avrithis, Ewa Kijak, Shashanka Venkataramanan.

Duration: 3.5 year, started in April 2020 Partners: Inria project-teams Linkmedia, CEA LIST, LIMSI, IRIT.

The overall goal of the project is to tackle the problem of ambiguities of visual and textual content by learning then combining their representations. As a final use case, we propose to solve a Multimedia Question Answering task, that requires to rely on three different sources of information to answer a (textual) question with regard to visual data as well as an external knowledge base containing millions of unique entities, each being represetd by textual and visual content as well as some links to other entities. An important work will deal with the representation of entities into a common tri-modal space, in which one should determine the content to associate to an entity to adequately represent it. The challenge consists in defining a representation that is compact (for performance) while still expressive enough to reflect the potential links between the entity and a variety of others.

MinArm: EVE3

Participants: Teddy Furon.

Duration: 3 year, started in April 2019 Partners: MinArm, CRIStAL Lille, LIRMM, Univ. Troyes, Univ. Paris Saclay

Teaching and technology survey on steganography and steganalysis in the real world.

ANR UNLIR: Unsupervised Representation Learning for Image Recognition

Participants: Yannis Avrithis.

Duration: 4 years, started in January 2020In relation with the JCJC awarded to Ronan Sicre, LIS, Aix-Marseille.

The project lies in the field of computer vision, pattern recognition, and machine learning. We study two problems of image recognition: image classification and image retrieval. Like machine learning, computer vision has witnessed a core change with the recent repopularization of Deep Neural Networks (DNN). Despite the success of DNN, several limitations are to be investigated.

1. Complex recognition problems such as fine grained classification (highly similar categories e.g. bird species, airplane/car models, etc.) show that state of the art DNNs are still improved by better objective functions and more discriminative intermediate representations.
2. Despite progress in using less annotated data, DNN can hardly cope with learning from few examples.
3. DNNs have so many parameters and complex structures that it is extremely hard to understand what happens in every layer in producing the final decision.

This project aims to address these limitations. In particular, we will work towards building networks capable of solving fine-grained visual recognition tasks. We will improve the capabilities of networks to learn from few to no data, building highly discriminative representations that can address complex recognition problems. Following that, we will provide insight on how such models take their decisions.

Computer vision for smart phones (MobilAI)

Participants: Yannis Avrithis, Mateusz Budnik.

Duration: 2 years, started in September 2018 Partners: Lamark, Quai des Apps, AriadNext

The ability of our mobile devices to process visual information is currently not limited by their camera or computing power but by the network. Many mobile apps suffer from long latency due to data transmitted over the network for visual search. MobilAI aims to provide fast visual recognition on mobile devices, offering quality user experience whatever the network conditions. The idea is to transfer efficient deep learning solutions for image classification and retrieval onto embedded platforms such as smart phones. The intention is to use such solutions in B2B and B2C application contexts, for instance recognizing products and ordering online, accessing information about artifacts in exhibitions, or identifying identity documents. In all cases, visual recognition is performed on the device, with minimal or no access to the network.

11.1.1 Scientific events: organisation

11.1.2 Scientific events: selection

11.1.3 Journal

11.1.4 Invited talks

• Teddy Furon at THALES internal seminar
• Teddy Furon at HUBDAY DS&IA, Pôle de Compétitivité Systematic

11.1.5 Leadership within the scientific community

• Laurent Amsaleg is a member of the Steering Committee of SISAP for the 2016-2020 term
• Laurent Amsaleg is a member of the Steering Committee of ACM Multimedia for the 2020-2023 term
• Vincent Claveau is a member of the Steering Committee of TALN conference (2018-2020)
• Guillaume Gravier is a member of the scientific committee of the GDR Traitement Automatique du Langage Naturel
• Guillaume Gravier is a member of Allistene's national network of AI referents
• Guillaume Gravier is a member of the steering committee of the graduate school (EUR) Digisport
• Guillaume Gravier is coordinating the AI doctoral program (ANR CDIA call) for Rennes
• Pascale Sébillot was a member of the permanent steering committee of Conf. francophone Traitement Automatique des Langues Naturelles (July 2013-July 2020).
• Pascale Sébillot is a member of the board of the pre-GDR Traitement Automatique des Langues; leader of the Intermodality and Multimodality working group till Dec. 2020.

11.1.6 Scientific expertise

• Vincent Claveau served as expert for the evaluation of DATAIA/MSH Paris-Saclay 'Projets excellence'
• Vincent Claveau served as expert for the evaluation of CIFRE PhD projects
• Guillaume Gravier served as an evaluator for the Belgian FED - tWIN program
• Guillaume Gravier evaluated tenure tracks application for UChile
• Guillaume Gravier was a member of the selection committee for INA's call for research projects
• Teddy Furon is Scientific advisor for the Imatag company
• Yannis Avrithis served as expert for the evaluation of Post-Doctoral Research projects for State Education Development Agency, Latvia
• Yannis Avrithis appointed as member of the Advisory Board of European H2020 project DeepCube

11.1.7 Research administration

• Guillaume Gravier is deputy director of IRISA
• Pascale Sébillot is the director of the Computer Science Laboratory, INSA Rennes, France.
• Pascale Sébillot is the deputy director of the Scientific Advisory Committee of IRISA UMR 6074.
• Pascale Sébillot is a member of the theses advisory committee of the MathSTIC doctoral school.
• Pascale Sébillot is a member of the board of the MathSTIC doctoral school.
• Teddy Furon is a member of Commission du Personnel Inria Rennes Bretagne Atlantique
• Laurent Amsaleg is a member of Commission formation permanente, Inria
• Laurent Amsaleg is a member of Commission des moyens incitatifs, Inria

11.2.1 Teaching

• Licence: Laurent Amsaleg, Bases de données avancées, 2h, L3-option génie mathématique, INSA Rennes, France
• Licence: Guillaume Gravier, Base de données, 26h, L2, INSA Rennes
• Licence: Guillaume Gravier, Natural language processing, 12h, L3, INSA Rennes
• Licence: Guillaume Gravier, Probability and statistics, 16h, L3, INSA Rennes, France
• Licence: Pascale Sébillot, Natural Language Processing, 10h, L3, INSA Rennes, France
• Licence: Simon Malinowski, Data Analysis, 32h, L3, ISTIC, Rennes, France
• Master: Laurent Amsaleg, Bases de données avancées, 25h, M2, INSA Rennes, France
• Engineering school: Vincent Claveau, Machine Learning, 18h, 3rd year, INSA Rennes, France
• Master: Vincent Claveau, Information Retrieval, 10h, M2 MIAGE, Univ. Rennes, France
• Master: Pascale Sébillot, Natural Language Processing, 6h, M1, INSA Rennes, France
• Master: Teddy Furon, Rare Event Simulations, INSA Rennes, France
• Master: Guillaume Gravier, Natural Language Processing, 6h, M1, INSA Rennes
• Master: Guillaume Gravier, Natural Language Processing, 21h, M2, ENSAI
• Master: Guillaume Gravier, Data analysis and probabilistic modeling, 30h, M2, Univ. Rennes 1
• Master: Ewa Kijak, Image processing, 55h, M1, ESIR, France
• Master: Ewa Kijak, Supervised machine learning, 15h, M2R, University Rennes 1, France
• Master: Ewa Kijak, Image classification, 45h, M1, ESIR, France
• Master: Ewa Kijak, Computer vision, 22h, M2, ESIR, France
• Engineering school: Vincent Claveau, Machine Learning, 18h, 3rd year, INSA Rennes, France
• Master: Vincent Claveau, Information Retrieval, 10h, M2 MIAGE, Univ. Rennes, France
• Master: Yannis Avrithis, Deep learning for vision, 20h, M2 SIF France
• Master: Yannis Avrithis, Computer vision, 30h, National and Kapodistrian University of Athens, Greece
• Master: Simon Malinowski, Basics of Data Analytics for Data Science, 24h, EIT Data Science Master 1, Rennes
• Master: Simon Malinowski, Prediction Methods, 30h, M1 MIAGE and Data Science EIT Master 1, Rennes
• Master: Simon Malinowski, Statisical Data Mining, 24h, M2 MIAGE, ISTIC, Rennes
• Master: Simon Malinowski, Symbolic Data Mining, 12h, M2 MIAGE, ISTIC, Rennes
• Simon Malinowski is responsible for the Master 2 MIAGE parcours Classique
• Simon Malinowski is responsible for the M2 studies within the DataScience track of the EIT-digital master school.

11.2.2 Supervision

• PhD in progress: Hanwei Zhang, Deep Learning in Adversarial Contexts, October 2017, Laurent Amsaleg, Yannis Avrithis, Teddy Furon
• PhD in progress: Yichang Wang, Adversarial methods for explainable time series classification. Started in April 2018. Simon Malinowski, Elisa Fromont, Romain Tavenard, Rémi Emonet.
• PhD in progress: Marzieh Gheisari Khorasgani, Secure identification in the Internet of Things, January 2018, Laurent Amsaleg & Teddy Furon
• PhD in progress: Antoine Perquin, Universal speech synthesis through embeddings of massive heterogeneous data, October 2017, Laurent Amsaleg, Gwénolé Lecorvé & Damien Lolive (with Expression, IRISA team)
• PhD in progress: Benoit Bonnet, Adversarial images, November 2019, Teddy Furon & Patrick Bas
• PhD in progress: Cheikh Brahim El Vaigh, Incremental content to data linking leveraging ontological knowledge in data journalism, started October 2017, Guillaume Gravier, Pascale Sébillot and François Goasdoué (with CEDAR, Inria team)
• PhD in progress: Cyrielle Mallart, Incremental dynamic construction of knowledge graphs from text mining, started December 2018, Guillaume Gravier, Michel Le Nouy (Ouest-France), Pascale Sébillot
• PhD in progress: Duc Hau Nguyen, Multimodal space for the generation and justification of semantic links between documents, started September 2020, Guillaume Gravier, Pascale Sébillot
• PhD in progress: François Torregrossa, Heterogeneous data embedding for professional search, started November 2018, Robin Allessiardo (So Local), Vincent Claveau, Guillaume Gravier
• PhD in progress: Yann Lifchitz, Few shot learning for object recognition in aerial images. Started March 2018, Yannis Avrithis & Sylvaine Picard (Safran Tech).
• PhD in progress: Raquel Almeida, Learning hierarchichal models for multimedia data, started January 2019, Ewa Kijak & Simon Malinowski & Laurent Amsaleg
• PhD in progress: Shashanka Venkataramanan, Metric learning for instance- and category-level visual representations. Started in December 2020. Yannis Avrithis, Ewa Kijak & Laurent Amsaleg
• PhD in progress: Thibault Maho, Black box attacks, Teddy Furon & Erwan Le Merrer
• PhD in progress: Samuel Tap, Deep learning in the encrypted domain, Teddy Furon
• PhD in progress: Karim Tit, Deep learning and uncertainties, Teddy Furon
• PhD in progress: Deniz Engin, Video query answering in domestic environments. Started in September 2020. Teddy Furon, Yannis Avrithis, Laurent Amsaleg
• PhD in progress: Antoine Chaffin, Multimodal detection of fake news, started November 2020, Ewa Kijak, Vincent Claveau
• PhD: Oriane Siméoni, Robust image representation for classification, retrieval and object discovery, defended Nov. 2020, Yannis Avrithis, Guillaume Gravier
• PhD: Colin Leverger, Probabilistic forecasting of seasonal time series, defended Nov. 2020, Simon Malinowski, Thomas Guyet, Laurence Rozé, Alexandre Termier
• PhD: Clément Dalloux, Text-mining and information extraction in clinical texts, Sup.: Vincent Claveau. Defended in Dec. 2020 43

11.2.3 Juries

• Vincent Claveau was reviewer for the mid-term PhD auditions of Ygor Gallina (LS2N)
• Vincent Claveau was reviewer for the PhD of Paul Mousset (Univ. Toulouse - IRIT)
• Vincent Claveau was reviewer for the PhD of Faneva Ramiandrisoa (Univ. Toulouse - IRIT)
• Guillaume Gravier was president of the HDR juries of Ngoc Quong (Univ. Rennes 1) and Aurelie Lemaitre (Univ. Rennes 2)
• Guillaume Gravier was a reviewer of the PhD thesis of Y. Le Gacheux, CNAM
• Pascale Sébillot was involved in the following juries:
  • HDR Peggy Cellier, Université Rennes 1, October 2020, member
  • HDR Richard Dufour, Avignon Université, December 2020, reviewer
• Teddy Furon was reviewer for the PhD of Alexandre Sablayrolles (Facebook - Inria Grenoble)
• Teddy Furon was a member of Comité de sélection for IUT St Dié, Univ. Lorraine
• Yannis Avrithis was involved in the following juries:
  • PhD Martin Engilberge, Paris-Sorbonne Université, June 2020, reviewer
  • PhD Patrick Bordes, Paris-Sorbonne Université, Nov. 2020, reviewer

11.3.1 Articles and contents

• interview for the journal 'Techniques pour l'ingénieur', March 2020
• Vincent Claveau: interview for the journal 'Sciences et vie'
• Vincent Claveau: interview for RFI Brazil, Feb. 2020
• Vincent Claveau: interview for Atlantico (Press), Oct. 2020

11.3.2 Education

• Vincent Claveau: Virtual conference/interview for students of Sciences Po Presse écrite about "Culture et enjeux du numérique"

11.3.3 Interventions

• Vincent Claveau: Public conference Table ronde "Vannes Territoire apprenant"

International journals

• 8 articleLaurentL. Amsaleg, JamesJ. Bailey, AmelieA. Barbe, SarahS. Erfani, TeddyT. Furon, MichaelM. Houle, MilosM. Radovanovic and Nguyen XuanN. Vinh. High Intrinsic Dimensionality Facilitates Adversarial Attack: Theoretical EvidenceIEEE Transactions on Information Forensics and Security16September 2020, 1-12
  HALDOIback to textback to text
• 9 article AlessioA. Antonini, MariM. Carmen Suárez-Figueroa, AlessandroA. Adamou, FrancescaF. Benatti, FrançoisF. Vignale, GuillaumeG. Gravier and LuciaL. Lupi. Understanding the phenomenology of reading through modelling Semantic Web – Interoperability, Usability, Applicability 2020
  HAL back to text
• 10 article KazukiK. Ashihara, Cheikh BrahimC. El Vaigh, ChenhuiC. Chu, BenjaminB. Renoust, NorikoN. Okubo, NorikoN. Takemura, YutaY. Nakashima and HajimeH. Nagahara. Improving topic modeling through homophily for legal documents Applied Network Science 5 1 December 2020
  HAL DOI back to text
• 11 articleGabrielG. Barbosa Da Fonseca, GabrielG. Sargent, RonanR. Sicre, ZeniltonZ. Kleber Gonçalves do Patrocinio, GuillaumeG. Gravier and Silvio Jamil F.S. Guimarães. Hierarchical Multi-Label Propagation using Speaking Face Graphs for Multimodal Person DiscoveryMultimedia Tools and Applications2020, 1-27
  HALDOIback to text
• 12 article Mihai GabrielM. Constantin, Liviu DanielL. Stefan, BogdanB. Ionescu, Claire-HeleneC.-H. Demarty, MatsM. Sjoberg, MarkusM. Schedl and GuillaumeG. Gravier. Affect in Multimedia: Benchmarking Violent Scenes Detection IEEE Transactions on Affective Computing 2020
  HAL DOI
• 13 article ClémentC. Dalloux, VincentV. Claveau, NataliaN. Grabar, Lucas Emanuel SilvaL. Oliveira, Claudia MariaC. Cabral Moro, Yohan BonesckiY. Gumiel and Deborah RibeiroD. Carvalho. Supervised learning for the detection of negation and of its scope in French and Brazilian Portuguese biomedical corpora Natural Language Engineering June 2020
  HAL DOI back to text
• 14 article NataliaN. Grabar, ClémentC. Dalloux and VincentV. Claveau. CAS: corpus of clinical cases in French Journal of Biomedical Semantics August 2020
  HAL DOI back to text
• 15 article FrançoisF. Torregrossa, RobinR. Allesiardo, VincentV. Claveau, NihelN. Kooli and GuillaumeG. Gravier. A survey on training and evaluation of word embeddings International Journal of Data Science and Analytics February 2021
  HAL DOI
• 16 article HanweiH. Zhang, YannisY. Avrithis, TeddyT. Furon and LaurentL. Amsaleg. Smooth adversarial examples EURASIP Journal on Information Security 2020 1 December 2020
  HAL DOI back to text
• 17 articleHanweiH. Zhang, YannisY. Avrithis, TeddyT. Furon and LaurentL. Amsaleg. Walking on the Edge: Fast, Low-Distortion Adversarial ExamplesIEEE Transactions on Information Forensics and Security16September 2020, 701 - 713
  HALDOIback to text

International peer-reviewed conferences

• 18 inproceedings BenoitB. Bonnet, TeddyT. Furon and PatrickP. Bas. Fooling an Automatic Image Quality Estimator MediaEval Benchmarking Intiative for Multimedia Evaluation (MediaEval 2020) Online, United States December 2020
  HAL back to text back to text
• 19 inproceedingsBenoîtB. Bonnet, TeddyT. Furon and PatrickP. Bas. What if Adversarial Samples were Digital Images?IH&MMSEC 2020 - 8th ACM Workshop on Information Hiding and Multimedia SecurityIH&MMSec '20: Proceedings of the 2020 ACM Workshop on Information Hiding and Multimedia SecurityDenver, FranceJune 2020, 1-11
  HALDOIback to textback to text
• 20 inproceedings VincentV. Claveau. Detecting fake news in tweets from text and propagation graph: IRISA's participation to the FakeNews task at MediaEval 2020 Proceedings of the MediaEval Benchmarking Initiative for Multimedia Evaluation (MediaEval 2020) MediaEval Benchmarking Initiative for Multimedia Evaluation (MediaEval 2020) online, United States December 2020
  HAL back to text back to text
• 21 inproceedingsVincentV. Claveau. Embedding medical concepts without textsActes de la 6e conférence conjointe Journées d'Études sur la Parole (JEP, 33e édition), Traitement Automatique des Langues Naturelles (TALN, 27e édition), Rencontre des Étudiants Chercheurs en Informatique pour le Traitement Automatique des Langues (RÉCITAL, 22e édition). Volume 2 : Traitement Automatique des Langues NaturellesJEP/TALN/RECITAL 2020 - 6e conférence conjointe Journées d'Études sur la Parole (JEP, 33e édition), Traitement Automatique des Langues Naturelles (TALN, 27e édition), Rencontre des Étudiants Chercheurs en Informatique pour le Traitement Automatique des Langues (RÉCITAL, 22e édition). Volume 2 : Traitement Automatique des Langues NaturellesNancy, FranceJune 2020, 181-188
  HALback to text
• 22 inproceedingsClémentC. Dalloux, VincentV. Claveau, MarcM. Cuggia, GuillaumeG. Bouzillé and NataliaN. Grabar. Supervised Learning for the ICD-10 Coding of French Clinical NarrativesMIE 2020 - Medical Informatics Europe conference - Digital Personalized Health and MedicineGeneva, SwitzerlandApril 2020, 1-5
  HALback to text
• 23 inproceedings Cheikh BrahimC. El Vaigh, FrançoisF. Goasdoué, GuillaumeG. Gravier and PascaleP. Sébillot. A Novel Path-based Entity Relatedness Measure for Efficient Collective Entity Linking International Semantic Web Conference (ISWC) Athens, Greece November 2020
  HAL back to text
• 24 inproceedings Cheikh BrahimC. El Vaigh, GuillaumeG. Le Noé-Bienvenu, GuillaumeG. Gravier and PascaleP. Sébillot. IRISA System for Entity Detection and Linking at CLEF HIPE 2020 CEUR Workshop Proceedings Thessaloniki, Greece September 2020
  HAL back to text
• 25 inproceedingsCheikh BrahimC. El Vaigh, FrançoisF. Torregrossa, RobinR. Allesiardo, GuillaumeG. Gravier and PascaleP. Sébillot. A correlation-based entity embedding approach for robust entity linkingICTAI 2020 - IEEE 32nd International Conference on Tools with Artificial IntelligenceVirtual, United StatesNovember 2020, 1-6
  HALback to text
• 26 inproceedings MarziehM. Gheisari, TeddyT. Furon and LaurentL. Amsaleg. Joint Learning of Assignment and Representation for Biometric Group Membership ICASSP 2020 - 45th International Conference on Acoustics, Speech, and Signal Processing Proc. of IEEE ICASSP Barcelona, Spain May 2020
  HAL back to text
• 27 inproceedingsAhmetA. Iscen, GiorgosG. Tolias, YannisY. Avrithis, OndřejO. Chum and CordeliaC. Schmid. Graph Convolutional Networks for Learning with Few Clean and Many Noisy LabelsECCV 2020 - 16th European Conference on Computer VisionVirtual, United KingdomNovember 2020, 286-302
  HALDOIback to text
• 28 inproceedings Suresh KirthiS. Kumaraswamy, MiaojingM. Shi and EwaE. Kijak. Detecting Human-Object Interaction with Mixed Supervision WACV 2021 - Winter Conference on Applications of Computer Vision Waikoloa / Virtual, United States January 2021
  HAL back to text
• 29 inproceedings YannY. Lifchitz, YannisY. Avrithis and SylvaineS. Picard. Few-Shot Few-Shot Learning and the role of Spatial Attention International Conference on Pattern Recognition Virtual, Italy January 2021
  HAL back to text
• 30 inproceedings YannY. Lifchitz, YannisY. Avrithis and SylvaineS. Picard. Local Propagation for Few-Shot Learning International Conference on Pattern Recognition Virtual, Italy January 2021
  HAL back to text
• 31 inproceedingsWenqingW. Liu, MiaojingM. Shi, TeddyT. Furon and LiL. Li. Defending Adversarial Examples via DNN Bottleneck ReinforcementACM Multimedia Conference 2020Proc. of ACM Multimedia ConferenceSeattle, United StatesOctober 2020, 1930-1938
  HALDOIback to text
• 32 inproceedings ZhuoranZ. Liu, ZhengyuZ. Zhao, MarthaM. Larson and LaurentL. Amsaleg. Exploring Quality Camouflage for Social Images MediaEval Benchmarking Initiative for Multimedia Evaluation (MediaEval 2020) Online, United States December 2020
  HAL back to text
• 33 inproceedingsCyrielleC. Mallart, MichelM. Le Nouy, GuillaumeG. Gravier and PascaleP. Sébillot. Relation, are you there? LSTM-based relation detection to improve knowledge extractionActes de la 6e conférence conjointe Journées d'Études sur la Parole (JEP, 33e édition), Traitement Automatique des Langues Naturelles (TALN, 27e édition), Rencontre des Étudiants Chercheurs en Informatique pour le Traitement Automatique des Langues (RÉCITAL, 22e édition). Volume 2 : Traitement Automatique des Langues NaturellesJEP-TALN-RECITAL 2020 - 6e conférence conjointe Journées d'Études sur la Parole (JEP, 33e édition), Traitement Automatique des Langues Naturelles (TALN, 27e édition), Rencontre des Étudiants Chercheurs en Informatique pour le Traitement Automatique des Langues (RÉCITAL, 22e édition). Volume 2 : Traitement Automatique des Langues NaturellesNancy, France2020, 279-287
  HALback to text
• 34 inproceedings OrianeO. Siméoni, MateuszM. Budnik, YannisY. Avrithis and GuillaumeG. Gravier. Rethinking deep active learning: Using unlabeled data at model training International Conference on Pattern Recognition Milan, Italy https://www.micc.unifi.it/icpr2020/ 2020
  HAL back to text
• 35 inproceedingsFrançoisF. Torregrossa, VincentV. Claveau, NihelN. Kooli, GuillaumeG. Gravier and RobinR. Allesiardo. On the Correlation of Word Embedding Evaluation MetricsLREC 2020 - 12th Conference on Language Resources and EvaluationMarseille, FranceMay 2020, 4789 - 4797
  HALback to text
• 36 inproceedingsFrançoisF. Torregrossa, GuillaumeG. Gravier, VincentV. Claveau and NihelN. Kooli. HierarX : un outil pour la découverte de hiérarchies dans des espaces hyperboliques à partir de similaritésEGC 2020 - 20ème Conférence sur l'Extraction et Gestion des ConnaissancesE-36Revue des Nouvelles Technologies de l'InformationBruxelles, Belgiumhttps://egc2020.sciencesconf.org/January 2020, 491 - 498
  HALback to text
• 37 inproceedings FrançoisF. Vignale, AlessioA. Antonini and GuillaumeG. Gravier. The Reading Experience Ontology (REO): Reusing and Extending CIDOC CRM Digital Humanities Digital Humanities Ottawa, Canada 2020
  HAL back to text
• 38 inproceedings YichangY. Wang, RémiR. Emonet, ElisaE. Fromont, SimonS. Malinowski and RomainR. Tavenard. Adversarial Regularization for Explainable-by-Design Time Series Classification ICTAI 2020 - 32th International Conference on Tools with Artificial Intelligence online, Greece November 2020
  HAL back to text

Scientific book chapters

• 39 inbookOmar ShahbazO. Khan, Björn ÞórB. Jónsson, StevanS. Rudinac, JanJ. Zahálka, HannaH. Ragnarsdóttir, ÞórhildurÞ. Þorleiksdóttir, Gylfi ÞórG. Guðmundsson, LaurentL. Amsaleg and MarcelM. Worring. Interactive Learning for Multimedia at LargeAdvances in Information Retrieval. ECIR 2020April 2020, 495-510
  HALDOIback to text

Edition (books, proceedings, special issue of a journal)

• 40 bookCécileC. Fabre, EmmanuelE. Morin, SophieS. Rosset and PascaleP. Sébillot. Varia - Préface - 60-1Traitement Automatique des Langues601www.atala.org/revuetalJanuary 2020, 7-11
  HAL

Doctoral dissertations and habilitation theses

• 41 thesis YannisY. Avrithis. Exploring and Learning from Visual Data Université de Rennes 1 July 2020
  HAL
• 42 thesis VincentV. Claveau. About Natural Language Processing for Information Retrieval and vice versa Univ. of Rennes January 2020
  HAL
• 43 thesis ClémentC. Dalloux. Text mining and information extraction in clinical data Université de Rennes 1 December 2020
  HAL back to text back to text back to text
• 44 thesis OrianeO. Siméoni. Robust image representation for classification, retrieval and object discovery Université rennes1 September 2020
  HAL

Reports & preprints

• 45 misc MateuszM. Budnik and YannisY. Avrithis. Asymmetric Metric Learning for Knowledge Transfer June 2020
  HAL back to text back to text
• 46 misc TeddyT. Furon. Note: An alternative proof of the vulnerability of $k$-NN classifiers in high intrinsic dimensionality regions' January 2021
  HAL back to text
• 47 misc MichalisM. Lazarou, YannisY. Avrithis and TaniaT. Stathaki. Iterative label cleaning for transductive and semi-supervised few-shot learning December 2020
  HAL back to text