Maxime Poli

I’m a first year PhD student in the Cognitive Machine Learning team of the LSCP at École Normale Supérieure, advised by Emmanuel Dupoux and Emmanuel Chemla.

I received an engineer’s degree from École des Ponts and a Master’s degree in Mathematics, Computer Vision and Machine Learning (MVA) from École Normale Supérieure Paris-Saclay.

News

Publications

Introducing topography in convolutional neural networks
Maxime Poli, Emmanuel Dupoux, and Rachid Riad.
ICASSP 2023
arXivdoicode abstract bibtex @inproceedings{poli2023introducing, title = {Introducing {{Topography}} in {{Convolutional Neural Networks}}}, booktitle = {{{ICASSP}} 2023 - 2023 {{IEEE International Conference}} on {{Acoustics}}, {{Speech}} and {{Signal Processing}} ({{ICASSP}})}, author = {Poli, Maxime and Dupoux, Emmanuel and Riad, Rachid}, year = {2023}, doi = {10.1109/ICASSP49357.2023.10096671}, }Parts of the brain that carry sensory tasks are organized topographically: nearby neurons are responsive to the same properties of input signals. Thus, in this work, inspired by the neuroscience literature, we proposed a new topographic inductive bias in Convolutional Neural Networks (CNNs). To achieve this, we introduced a new topographic loss and an efficient implementation to topographically organize each convolutional layer of any CNN. We benchmarked our new method on 4 datasets and 3 models in vision and audio tasks and showed equivalent performance to all benchmarks. Besides, we also showcased the generalizability of our topographic loss with how it can be used with different topographic organizations in CNNs. Finally, we demonstrated that adding the topographic inductive bias made CNNs more resistant to pruning. Our approach provides a new avenue to obtain models that are more memory efficient while maintaining better accuracy.

Shennong: a Python toolbox for audio speech features extraction
Mathieu Bernard*, Maxime Poli*, Julien Karadayi, and Emmanuel Dupoux.
Behavior Research Methods
arXivdoicode abstract bibtex @article{bernard2023shennong, title = {Shennong: {{A Python}} Toolbox for Audio Speech Features Extraction}, author = {Bernard, Mathieu and Poli, Maxime and Karadayi, Julien and Dupoux, Emmanuel}, year = {2023}, journal = {Behavior Research Methods}, doi = {10.3758/s13428-022-02029-6}, }We introduce Shennong, a Python toolbox and command-line utility for audio speech features extraction. It implements a wide range of well-established state-of-the-art algorithms: spectro-temporal filters such as Mel-Frequency Cepstral Filterbank or Predictive Linear Filters, pre-trained neural networks, pitch estimators, speaker normalization methods, and post-processing algorithms. Shennong is an open source, reliable and extensible framework built on top of the popular Kaldi speech processing library. The Python implementation makes it easy to use by non-technical users and integrates with thirdparty speech modeling and machine learning tools from the Python ecosystem. This paper describes the Shennong software architecture, its core components, and implemented algorithms. Then, three applications illustrate its use. We first present a benchmark of speech features extraction algorithms available in Shennong on a phone discrimination task. We then analyze the performances of a speaker normalization model as a function of the speech duration used for training. We finally compare pitch estimation algorithms on speech under various noise conditions.