Search

Recent progress in natural language processing has been adapted to the symbolic music modality. Language models, such as Transformers, have been used with symbolic music for a variety of tasks among which music generation, modeling or transcription, with state-of-the-art performances. These models are beginning to be used in production products. To encode and decode music for the backbone model, they need to rely on tokenizers, whose role is to serialize music into sequences of distinct elements called tokens. MidiTok is an open-source library allowing to tokenize symbolic music with great flexibility and extended features. It features the most popular music tokenizations, under a unified API. It is made to be easily used and extensible for everyone., Comment: Updated and comprehensive report. Original ISMIR 2021 document at https://archives.ismir.net/ismir2021/latebreaking/000005.pdf

Symbolic music is widely used in various deep learning tasks, including generation, transcription, synthesis, and Music Information Retrieval (MIR). It is mostly employed with discrete models like Transformers, which require music to be tokenized, i.e., formatted into sequences of distinct elements called tokens. Tokenization can be performed in different ways. As Transformer can struggle at reasoning, but capture more easily explicit information, it is important to study how the way the information is represented for such model impact their performances. In this work, we analyze the common tokenization methods and experiment with time and note duration representations. We compare the performances of these two impactful criteria on several tasks, including composer and emotion classification, music generation, and sequence representation learning. We demonstrate that explicit information leads to better results depending on the task., Comment: ISMIR 2023

The issue of civil society participation in the processes of developing, implementing, and monitoring national climate plans is crucial to ensuring that these ambitions are both acceptable to the populations and, beyond that, that they align with a trajectory of sustainable development beneficial to all actors in Senegalese society. This case study aims to analyze the degree of involvement through consultations with a variety of civil society actors, as well as members of the administration and international partners active in the fight against climate change. While some believe that civil society participation has progressed significantly in recent years, the vast majority feel that the level reached is still insufficient., Barriers to civil society’s appropriation of climate issues include, among others, the lack of representativeness of grassroots organizations and vulnerable groups, as well as the unfamiliarity of civil society organizations (CSOs) and community-based organizations (CBOs) with the processes, objectives, and stakeholders of the National Climate Plan (NCP), meaning that many civil society actors are rarely aware of their contribution to its implementation even though they are involved. The study also shows how limited human and financial resources reduce the possibility of broad participation by civil society organizations, especially outside the capital. This raises the question of a fair, adequate, and targeted allocation of climate financing to meet the ambitions of climate policies in Senegal. This report presents recommendations to overcome barriers that may explain low ownership of national climate plans by communities, in order to propose ways for populations to be key actors in an ambitious ecological transition in Senegal., La question de la participation de la société civile aux processus d’élaboration, de mise en oeuvre et de suivi des plans nationaux sur le climat est clé afin de s’assurer que ces ambitions soient à la fois acceptables pour les populations et, au-delà, qu’elles s’inscrivent dans une trajectoire de développement durable bénéfique à l’ensemble des acteurs de la société sénégalaise., Cette étude de cas vise à analyser ce degré d’implication, à travers des consultations menées auprès d’une variété d’acteurs de la société civile, ainsi que des membres de l’administration et des partenaires internationaux actifs dans la lutte contre les changements climatiques. Alors que certains estiment que la participation de la société civile a beaucoup progressé dans les dernières années, la large majorité juge que le niveau atteint est encore insuffisant., Les barrières à l’appropriation des enjeux climatiques par la société civile incluent, entre autres, le manque de représentativité des organisations communautaires de base et des groupes vulnérables, mais également la méconnaissance des processus, objectifs et interlocuteurs de la CDN de la part des OSC et des OCB, impliquant que de nombreux acteurs de la société civile sont rarement conscients de contribuer à sa mise en oeuvre même si cela est le cas., L’étude montre également comment des ressources humaines et financières limitées réduisent la possibilité d’une large participation des organisations de la société civile, en particulier en dehors de la capitale. Cela soulève la question d’une allocation juste, adéquate et ciblée des financements climatiques pour répondre aux ambitions des politiques climatiques au Sénégal. Ce rapport présente des recommandations afin de surmonter les barrières qui expliqueraient une faible appropriation des plans nationaux sur le climat par les communautés, afin de proposer des voies pour que les populations soient des acteurs au coeur d’une transition écologique ambitieuse au Sénégal.

When used with deep learning, the symbolic music modality is often coupled with language model architectures. To do so, the music needs to be tokenized, i.e. converted into a sequence of discrete tokens. This can be achieved by different approaches, as music can be composed of simultaneous tracks, of simultaneous notes with several attributes. Until now, the proposed tokenizations rely on small vocabularies of tokens describing the note attributes and time events, resulting in fairly long token sequences, and a sub-optimal use of the embedding space of language models. Recent research has put efforts on reducing the overall sequence length by merging embeddings or combining tokens. In this paper, we show that Byte Pair Encoding, a compression technique widely used for natural language, significantly decreases the sequence length while increasing the vocabulary size. By doing so, we leverage the embedding capabilities of such models with more expressive tokens, resulting in both better results and faster inference in generation and classification tasks. The source code is shared on Github, along with a companion website. Finally, BPE is directly implemented in MidiTok, allowing the reader to easily benefit from this method., Comment: EMNLP 2023, source code: https://github.com/Natooz/BPE-Symbolic-Music

The Commission Femmes et Astronomie conducted a statistical study that aims at mapping the presence of women in French professional Astronomy today, and set a starting point for studying its evolution with time. For the year 2021, we proceeded with a sub-set of 8 astronomy and astrophysics institutes, hosting a total of 1060 employees, among which PhD students, post-doctoral researchers, and academic, technical, and administrative staff, representing around 25% of the community. We have investigated how the percentage of women vary with career stage, level of responsibility, job security, and level of income. The results of this preliminary study seem to illustrate the leaky pipeline, with one major bottleneck being the access to permanent positions. It appears that the proportion of women steadily decreases with the security of jobs, with the career stage, with the qualification level and with the income level., Comment: 6 pages, 5 figures

Recent works on predictive uncertainty estimation have shown promising results on Out-Of-Distribution (OOD) detection for semantic segmentation. However, these methods struggle to precisely locate the point of interest in the image, i.e, the anomaly. This limitation is due to the difficulty of finegrained prediction at the pixel level. To address this issue, we build upon the recent ObsNet approach by providing object instance knowledge to the observer. We extend ObsNet by harnessing an instance-wise mask prediction. We use an additional, class agnostic, object detector to filter and aggregate observer predictions. Finally, we predict an unique anomaly score for each instance in the image. We show that our proposed method accurately disentangles in-distribution objects from OOD objects on three datasets.

This paper presents an architecture for generating music for video games based on the Transformer deep learning model. Our motivation is to be able to customize the generation according to the taste of the player, who can select a corpus of training examples, corresponding to his preferred musical style. The system generates various musical layers, following the standard layering strategy currently used by composers designing video game music. To adapt the music generated to the game play and to the player(s) situation, we are using an arousal-valence model of emotions, in order to control the selection of musical layers. We discuss current limitations and prospects for the future, such as collaborative and interactive control of the musical components.

The objective of this chapter is to propose some retrospective analysis of the evolution of programming abstractions, from {\em procedures}, {\em objects}, {\em actors}, {\em components}, {\em services}, up to {\em agents}, %have some compare concepts of software component and of agent (and multi-agent system), %The method chosen is to by replacing them within a general historical perspective. Some common referential with three axes/dimensions is chosen: {\em action selection} at the level of one entity, {\em coupling flexibility} between entities, and {\em abstraction level}. We indeed may observe some continuous quest for higher flexibility (through notions such as {\em late binding}, or {\em reification} of {\em connections}) and higher level of {\em abstraction}. Concepts of components, services and agents have some common objectives (notably, {\em software modularity and reconfigurability}), with multi-agent systems raising further concepts of {\em autonomy} and {\em coordination}. notably through the notion of {\em auto-organization} and the use of {\em knowledge}. We hope that this analysis helps at highlighting some of the basic forces motivating the progress of programming abstractions and therefore that it may provide some seeds for the reflection about future programming abstractions., Comment: This preprint has been published as a chapter of a book about the French school of programming, coordinated by Bertrand Meyer and published by Springer in 2024

In this paper, we tackle the detection of out-of-distribution (OOD) objects in semantic segmentation. By analyzing the literature, we found that current methods are either accurate or fast but not both which limits their usability in real world applications. To get the best of both aspects, we propose to mitigate the common shortcomings by following four design principles: decoupling the OOD detection from the segmentation task, observing the entire segmentation network instead of just its output, generating training data for the OOD detector by leveraging blind spots in the segmentation network and focusing the generated data on localized regions in the image to simulate OOD objects. Our main contribution is a new OOD detection architecture called ObsNet associated with a dedicated training scheme based on Local Adversarial Attacks (LAA). We validate the soundness of our approach across numerous ablation studies. We also show it obtains top performances both in speed and accuracy when compared to ten recent methods of the literature on three different datasets.

This paper presents a comparative analysis on two artificial neural networks (with different architectures) for the task of tempo estimation. For this purpose, it also proposes the modeling, training and evaluation of a B-RNN (Bidirectional Recurrent Neural Network) model capable of estimating tempo in bpm (beats per minutes) of musical pieces, without using external auxiliary modules. An extensive database (12,550 pieces in total) was curated to conduct a quantitative and qualitative analysis over the experiment. Percussion-only tracks were also included in the dataset. The performance of the B-RNN is compared to that of state-of-the-art models. For further comparison, a state-of-the-art CNN was also retrained with the same datasets used for the B-RNN training. Evaluation results for each model and datasets are presented and discussed, as well as observations and ideas for future research. Tempo estimation was more accurate for the percussion only dataset, suggesting that the estimation can be more accurate for percussion-only tracks, although further experiments (with more of such datasets) should be made to gather stronger evidence.

In this paper we address the collective management of environmental commons with multiple usages in the framework of the mathematical viability theory. We consider that the stakeholders can derive from the study of their own socioeconomic problem the variables describing their different usages of the commons and its evolution, and a representation of the desirable states for the commons. We then consider the guaranteed viability kernel, subset of the set of desirable states where it is possible to maintain the state of the commons even when its evolution is represented by several conflicting models. This approach is illustrated on a problem of lake eutrophication., Comment: 25 pages, 5 figures Viability theory reminder in appendix, a diagram and references added

In most scientific domains, the deep learning community has largely focused on the quality of deep generative models, resulting in highly accurate and successful solutions. However, this race for quality comes at a tremendous computational cost, which incurs vast energy consumption and greenhouse gas emissions. At the heart of this problem are the measures that we use as a scientific community to evaluate our work. In this paper, we suggest relying on a multi-objective measure based on Pareto optimality, which takes into account both the quality of the model and its energy consumption. By applying our measure on the current state-of-the-art in generative audio models, we show that it can drastically change the significance of the results. We believe that this type of metric can be widely used by the community to evaluate their work, while putting computational cost -- and in fine energy consumption -- in the spotlight of deep learning research., Comment: 5 pages, 2 figures, ICASSP 2022

In this paper, we show how uncertainty estimation can be leveraged to enable safety critical image segmentation in autonomous driving, by triggering a fallback behavior if a target accuracy cannot be guaranteed. We introduce a new uncertainty measure based on disagreeing predictions as measured by a dissimilarity function. We propose to estimate this dissimilarity by training a deep neural architecture in parallel to the task-specific network. It allows this observer to be dedicated to the uncertainty estimation, and let the task-specific network make predictions. We propose to use self-supervision to train the observer, which implies that our method does not require additional training data. We show experimentally that our proposed approach is much less computationally intensive at inference time than competing methods (e.g. MCDropout), while delivering better results on safety-oriented evaluation metrics on the CamVid dataset, especially in the case of glare artifacts.

The current wave of deep learning (the hyper-vitamined return of artificial neural networks) applies not only to traditional statistical machine learning tasks: prediction and classification (e.g., for weather prediction and pattern recognition), but has already conquered other areas, such as translation. A growing area of application is the generation of creative content, notably the case of music, the topic of this paper. The motivation is in using the capacity of modern deep learning techniques to automatically learn musical styles from arbitrary musical corpora and then to generate musical samples from the estimated distribution, with some degree of control over the generation. This paper provides a tutorial on music generation based on deep learning techniques. After a short introduction to the topic illustrated by a recent exemple, the paper analyzes some early works from the late 1980s using artificial neural networks for music generation and how their pioneering contributions have prefigured current techniques. Then, we introduce some conceptual framework to analyze the various concepts and dimensions involved. Various examples of recent systems are introduced and analyzed to illustrate the variety of concerns and of techniques., Comment: To appear in the Special Issue on Art, Sound and Design in the Neural Computing and Applications Journal

State of the art Deep Neural Networks (DNN) can now achieve above human level accuracy on image classification tasks. However their outstanding performances come along with a complex inference mechanism making them arduously interpretable models. In order to understand the underlying prediction rules of DNNs, Dhamdhere et al. propose an interpretability method to break down a DNN prediction score as sum of its hidden unit contributions, in the form of a metric called conductance. Analyzing conductances of DNN hidden units, we find out there is a difference in how wrong and correct predictions are inferred. We identify distinguishable patterns of hidden unit activations for wrong and correct predictions. We then use an error detector in the form of a binary classifier on top of the DNN to automatically discriminate wrong and correct predictions of the DNN based on their hidden unit activations. Detected wrong predictions are discarded, increasing the model robustness. A different approach to distinguish wrong and correct predictions of DNNs is proposed by Wang et al. whose method is based on the premise that input samples leading a DNN into making wrong predictions are less stable to the DNN weight changes than correctly classified input samples. In our study, we compare both methods and find out by combining them that better detection of wrong predictions can be achieved., Comment: 2019 ICCV workshop on Interpreting and Explaining Visual AI models; 8 pages

We demonstrate the successive appearance of the exciton, biexciton, and P band of the exciton-exciton scattering with increasing excitation power in the photoluminescence of indium selenide layered crystals. The strict energy and momentum conservation rules of the P band are used to reexamine the exciton binding energy. The new value $\geq 20$ meV is markedly higher than the currently accepted 14 meV, being however well consistent with the robustness of excitons up to room temperature. A peak controlled by the Sommerfeld factor is found near the bandgap ($\sim 1.36$ eV), which puts the question on the pure three-dimensional character of the exciton in InSe, which has been assumed up to now. Our findings are of paramount importance for the successful application of InSe in nanophotonics., Comment: 6 pages, 4 figures

Nowadays the more powerful method to detect extrasolar planets is the transit method. We review the planet transits which were anticipated, searched, and the first ones which were observed all through history. Indeed transits of planets in front of their star were first investigated and studied in the solar system. The first observations of sunspots were sometimes mistaken for transits of unknown planets. The first scientific observation and study of a transit in the solar system was the observation of Mercury transit by Pierre Gassendi in 1631. Because observations of Venus transits could give a way to determine the distance Sun-Earth, transits of Venus were overwhelmingly observed. Some objects which actually do not exist were searched by their hypothetical transits on the Sun, as some examples a Venus satellite and an infra-mercurial planet. We evoke the possibly first use of the hypothesis of an exoplanet transit to explain some periodic variations of the luminosity of a star, namely the star Algol, during the eighteen century. Then we review the predictions of detection of exoplanets by their transits, those predictions being sometimes ancient, and made by astronomers as well as popular science writers. However, these very interesting predictions were never published in peer-reviewed journals specialized in astronomical discoveries and results. A possible transit of the planet beta Pic b was observed in 1981. Shall we see another transit expected for the same planet during 2018? Today, some studies of transits which are connected to hypothetical extraterrestrial civilisations are published in astronomical refereed journals. Some studies which would be classified not long ago as science fiction are now considered as scientific ones., Comment: Submiited to Handbook of Exoplanets (Springer)

In addition to traditional tasks such as prediction, classification and translation, deep learning is receiving growing attention as an approach for music generation, as witnessed by recent research groups such as Magenta at Google and CTRL (Creator Technology Research Lab) at Spotify. The motivation is in using the capacity of deep learning architectures and training techniques to automatically learn musical styles from arbitrary musical corpora and then to generate samples from the estimated distribution. However, a direct application of deep learning to generate content rapidly reaches limits as the generated content tends to mimic the training set without exhibiting true creativity. Moreover, deep learning architectures do not offer direct ways for controlling generation (e.g., imposing some tonality or other arbitrary constraints). Furthermore, deep learning architectures alone are autistic automata which generate music autonomously without human user interaction, far from the objective of interactively assisting musicians to compose and refine music. Issues such as: control, structure, creativity and interactivity are the focus of our analysis. In this paper, we select some limitations of a direct application of deep learning to music generation, analyze why the issues are not fulfilled and how to address them by possible approaches. Various examples of recent systems are cited as examples of promising directions., Comment: 17 pages. arXiv admin note: substantial text overlap with arXiv:1709.01620. Accepted for publication in Special Issue on Deep learning for music and audio, Neural Computing & Applications, Springer Nature, 2018

This paper is a survey and an analysis of different ways of using deep learning (deep artificial neural networks) to generate musical content. We propose a methodology based on five dimensions for our analysis: Objective - What musical content is to be generated? Examples are: melody, polyphony, accompaniment or counterpoint. - For what destination and for what use? To be performed by a human(s) (in the case of a musical score), or by a machine (in the case of an audio file). Representation - What are the concepts to be manipulated? Examples are: waveform, spectrogram, note, chord, meter and beat. - What format is to be used? Examples are: MIDI, piano roll or text. - How will the representation be encoded? Examples are: scalar, one-hot or many-hot. Architecture - What type(s) of deep neural network is (are) to be used? Examples are: feedforward network, recurrent network, autoencoder or generative adversarial networks. Challenge - What are the limitations and open challenges? Examples are: variability, interactivity and creativity. Strategy - How do we model and control the process of generation? Examples are: single-step feedforward, iterative feedforward, sampling or input manipulation. For each dimension, we conduct a comparative analysis of various models and techniques and we propose some tentative multidimensional typology. This typology is bottom-up, based on the analysis of many existing deep-learning based systems for music generation selected from the relevant literature. These systems are described and are used to exemplify the various choices of objective, representation, architecture, challenge and strategy. The last section includes some discussion and some prospects., Comment: 209 pages. This paper is a simplified version of the book: J.-P. Briot, G. Hadjeres and F.-D. Pachet, Deep Learning Techniques for Music Generation, Computational Synthesis and Creative Systems, Springer, 2019