Filters








2,679 Hits in 6.4 sec

An Efficient Posterior Regularized Latent Variable Model for Interactive Sound Source Separation

Nicholas J. Bryan, Gautham J. Mysore
2013 International Conference on Machine Learning  
To overcome these issues, we allow a user to interactively constrain a latent variable model by painting on a time-frequency display of sound to guide the learning process.  ...  allowing for high-quality interactive-rate separation without explicit training data.  ...  Bryan was an intern at Adobe Research.  ... 
dblp:conf/icml/BryanM13 fatcat:5gzihmivsbbvrpdnk2en4pjita

Speech Enhancement Using an Iterative Posterior NMF [chapter]

Sunnydayal Vanambathina
2019 New Frontiers in Brain-Computer Interfaces [Working Title]  
A speech enhancement method based on regularized nonnegative matrix factorization (NMF) for nonstationary Gaussian noise is proposed.  ...  The spectral components of speech and noise are modeled as Gamma and Rayleigh, respectively.  ...  carried out in single-channel sound source separation methods.  ... 
doi:10.5772/intechopen.84976 fatcat:3iwcqaim2nedhfpilpn4te2lke

Assisted Sound Sample Generation with Musical Conditioning in Adversarial Auto-Encoders [article]

Adrien Bitton, Philippe Esling, Antoine Caillon, Martin Fouilleul
2019 arXiv   pre-print
We condition the decoder for control over the rendered note attributes and use latent adversarial training for learning expressive style parameters that can ultimately be mixed.  ...  Our solution remains rather fast to train, it can directly be applied to other sound domains, including an user's libraries with custom sound tags that could be mapped to specific generative controls.  ...  Our goal is to learn expressive style variables from any sound tags, so that the model fosters creativity and assists digital interactions in music production.  ... 
arXiv:1904.06215v2 fatcat:weqemk2dtvgdrlhrb5bzzm3dui

A Survey on Probabilistic Models in Human Perception and Machines

Lux Li, Robert Rehr, Patrick Bruns, Timo Gerkmann, Brigitte Röder
2020 Frontiers in Robotics and AI  
However, little true integration between these fields exists in their applications of the probabilistic models for solving analogous problems, such as noise reduction, signal enhancement, and source separation  ...  We focus on audio and audio-visual processing, using examples of speech enhancement, automatic speech recognition, audio-visual cue integration, source separation, and causal inference to illustrate the  ...  as a latent variable S.  ... 
doi:10.3389/frobt.2020.00085 pmid:33501252 pmcid:PMC7805657 fatcat:oyuo7rdz4jcntkzt5rvgeu5deu

Representation Learning of Resting State fMRI with Variational Autoencoder [article]

Jung-Hoon Kim, Yizhen Zhang, Kuan Han, Minkyu Choi, Zhongming Liu
2020 bioRxiv   pre-print
After being trained with large data from the Human Connectome Project, the model has learned to represent and generate patterns of cortical activity and connectivity using latent variables.  ...  Here we establish a variational auto-encoder, as a generative model trainable with unsupervised learning, to disentangle the unknown sources of rs-fMRI activity.  ...  In the VAE model, the sources are the latent variables; the decoder describes how the sources generate the observed activity; the encoder models the inverse inference of the sources from the activity.  ... 
doi:10.1101/2020.06.16.155937 fatcat:bnuq7hcj3bdwtphu4vcthg4l6i

Cross-modal variational inference for bijective signal-symbol translation [article]

Axel Chemla–Romeu-Santos, Stavros Ntalampiras, Philippe Esling, Goffredo Haus, Gérard Assayag
2020 arXiv   pre-print
variables.  ...  We estimate this joint distribution with two different variational auto-encoders, one for each domain, whose inner representations are forced to match with an additive constraint, allowing both models  ...  Finally, we bind our transcription approach with a novel source-separation approach, based on explicit source decomposition with disjoint decoders.  ... 
arXiv:2002.03862v1 fatcat:4nowcusvd5g2xgeanls7poci7a

A Variational EM Algorithm for the Separation of Time-Varying Convolutive Audio Mixtures

Dionyssos Kounades-Bastian, Laurent Girin, Xavier Alameda-Pineda, Sharon Gannot, Radu Horaud
2016 IEEE/ACM Transactions on Audio Speech and Language Processing  
The sound sources are then separated by Wiener filters constructed with the estimators provided by the VEM algorithm.  ...  This paper addresses the problem of separating audio sources from time-varying convolutive mixtures.  ...  proposed continuous latent model, thus using localization cues to help the automatic separation of sound sources.  ... 
doi:10.1109/taslp.2016.2554286 fatcat:aikgw6nh6ncklmcirc73pq63mu

Latent Variable Algorithms for Multimodal Learning and Sensor Fusion [article]

Lijiang Guo
2019 arXiv   pre-print
We study multimodal learning and sensor fusion from a latent variable perspective. We first present a regularized recurrent attention filter for sensor fusion.  ...  We propose a co-learning approach using probabilistic graphical model which imposes a structural prior on the generative model: multimodal variational RNN (MVRNN) model, and derive a variational lower  ...  Lantao Liu for helpful discussions.  ... 
arXiv:1904.10450v1 fatcat:6634ghs74fcd3fz3l4nov4rb3m

Application of Bayesian structural equation modeling for examining phytoplankton dynamics in the Neuse River Estuary (North Carolina, USA)

G.B. Arhonditsis, H.W. Paerl, L.M. Valdes-Weaver, C.A. Stow, L.J. Steinberg, K.H. Reckhow
2007 Estuarine, Coastal and Shelf Science  
Finally, the optimal down-estuary grouping aggregates diatoms and chlorophytes, lumps together dinoflagellates with cryptophytes, while cyanobacteria are treated separately.  ...  We introduce a Bayesian structural equation modeling framework to explore the spatiotemporal phytoplankton community patterns in the Neuse River Estuary (study period 1995e2001).  ...  The latter error source reflects the latent variable model efficiency, i.e., ''how well can the physical environment, nitrogen, and temperature describe phytoplankton''.  ... 
doi:10.1016/j.ecss.2006.09.022 fatcat:z5nhczbrp5fqhgiufvvjpbjccm

From neural PCA to deep unsupervised learning [article]

Harri Valpola
2015 arXiv   pre-print
While standard autoencoders are analogous to latent variable models with a single layer of stochastic variables, the proposed network is analogous to hierarchical latent variables models.  ...  Learning combines denoising autoencoder and denoising sources separation frameworks.  ...  Acknowledgments I would like to thank Tapani Raiko, Antti Rasmus and Yoshua Bengio for useful discussions.  ... 
arXiv:1411.7783v2 fatcat:jfar5jv5xjarpfpmk4zsgmuui4

STRFs in primary auditory cortex emerge from masking-based statistics of natural sounds

Abdul-Saboor Sheikh, Nicol S. Harper, Jakob Drefs, Yosef Singer, Zhenwen Dai, Richard E. Turner, Jörg Lücke, Frédéric E. Theunissen
2019 PLoS Computational Biology  
An efficient truncated EM algorithm is used to fit the MCA model to cochleagram data.  ...  We therefore consider a new encoding approach for natural sounds, which combines a model of early auditory processing with maximal causes analysis (MCA), a sparse coding model which captures both the non-linear  ...  Such non-linear interactions give rise to psychoacoustic masking effects, which have been successfully exploited in technical applications such as source separation (e.g., [15] [16] [17] ).  ... 
doi:10.1371/journal.pcbi.1006595 pmid:30653497 pmcid:PMC6382252 fatcat:zy6szk5w4fcatlwe6icetutdcq

An Introduction to Variational Autoencoders

Diederik P. Kingma, Max Welling
2019 Foundations and Trends® in Machine Learning  
Variational autoencoders provide a principled framework for learning deep latent-variable models and corresponding inference models.  ...  In this work, we provide an introduction to variational autoencoders and some important extensions.  ...  Acknowledgements We are grateful for the help of Tim Salimans, Alec Radford, Rif A. Saurous and others who have given us valuable feedback at various stages of writing.  ... 
doi:10.1561/2200000056 fatcat:t3x7k3dt65a5rlviyiixdnj3yi

Deep Latent-Variable Models for Text Generation [article]

Xiaoyu Shen
2022 arXiv   pre-print
This dissertation presents how deep latent-variable models can improve over the standard encoder-decoder model for text generation.  ...  Deep latent-variable models, by specifying the probabilistic distribution over an intermediate latent process, provide a potential way of addressing these problems while maintaining the expressive power  ...  For the latter direction, we need to use a more flexible prior or posterior distribution for latent variables.  ... 
arXiv:2203.02055v1 fatcat:sq3upxl7xvfnhigoc7apszomwu

Deep generative models for musical audio synthesis [article]

M. Huzaifah, L. Wyse
2020 arXiv   pre-print
Sound modelling is the process of developing algorithms that generate sound under parametric control.  ...  While each of these approaches has been able to achieve high-quality synthesis and interaction for specific applications, they are all labour-intensive and each comes with its own challenges for designing  ...  Acknowledgements This research was supported by a Singapore MOE Tier 2 grant, "Learning Generative Recurrent Neural Networks," and by an NVIDIA Corporation Academic Programs GPU grant.  ... 
arXiv:2006.06426v2 fatcat:swt7npt3gnbj5ppzcf2ef3rose

Learning Stochastic Inverses

Andreas Stuhlmüller, Jessica Taylor, Noah D. Goodman
2013 Neural Information Processing Systems  
We explore the efficiency of this sampler for a variety of parameter regimes and Bayes nets.  ...  We show that estimated inverses converge asymptotically in number of (prior or posterior) training samples.  ...  Acknowledgments We thank Ramki Gummadi and anonymous reviewers for useful comments. This work was supported by a John S. McDonnell Foundation Scholar Award.  ... 
dblp:conf/nips/StuhlmullerTG13 fatcat:l6nhngbdmnhazkjp5r7xl4txjy
« Previous Showing results 1 — 15 out of 2,679 results