Weakly Supervised Audio Source Separation via Spectrum Energy Preserved Wasserstein Learning

Zhang, Ning; Yan, Junchi; Zhou, Yuchen

doi:10.48550/arxiv.1711.04121

Cited by 2 publications

(2 citation statements)

References 16 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We use a conditional GAN 38 , as the generative model in our generative decoder (for ease of presentation, we abuse a bit the notation of G(•) to also denote the generative model), which is adopted from the implementation of 44 and pre-trained using digits in the cGAN dataset. Note that this is the only supervision we have in this task, which is even weaker than the general concept of the weakly-supervised setting 45 . Given the shape (z i,j ) and probability embeddings (P i and Q i ), DRNets estimate the two digits in the cell by computing the expected digits over P i and Q i , i.e., n j=1 P i,j G(z i,j ) and n j=1 Q i,j G(z i,j+4 ), and remix them to reconstruct the original input mixture (Fig.…”

Section: Competing Interestsmentioning

confidence: 99%

Automating Crystal-Structure Phase Mapping: Combining Deep Learning with Constraint Reasoning

Chen¹,

Bai²,

Ament³

et al. 2021

Preprint

View full text Add to dashboard Cite

Crystal-structure phase mapping is a core, long-standing challenge in materials science that requires identifying crystal structures, or mixtures thereof, in synthesized materials. Materials science experts excel at solving simple systems but cannot solve complex systems, creating a major bottleneck in high-throughput materials discovery. Herein we show how to automate crystal-structure phase mapping. We formulate phase mapping as an unsupervised pattern demixing problem and describe how to solve it using Deep Reasoning Networks (DRNets). DRNets combine deep learning with constraint reasoning for incorporating scientific prior knowledge and consequently require only a modest amount of (unlabeled) data. DRNets compensate for the limited data by exploiting and magnifying the rich prior-knowledge about the thermodynamic rules governing the mixtures of crystals with constraint reasoning seamlessly integrated into neural network optimization. DRNets are designed with an interpretable latent space for encoding prior-knowledge domain constraints and seamlessly integrate constraint reasoning into neural network optimization. DRNets surpass previous approaches on crystal-structure phase mapping, unraveling the Bi-Cu-V oxide phase diagram, and aiding the discovery of solar-fuels materials.Artificial Intelligence (AI) 1 aims to develop intelligent systems, inspired in part by human intelligence. AI systems are now performing at human and even superhuman levels on a range of tasks such as image identification 2 , face, 3 and speech recognition 4 . AI also has the potential to accelerate scientific discovery dramatically. [5][6][7][8][9][10] Recent AI achievements have been driven mainly by advances in supervised deep learning 11 , which requires large labeled datasets to supervise model training. However, in general, scientists do not have large amounts of labeled data for scientific discovery. They often solve complex tasks using only a few data samples by amplifying intuitive pattern recognition with detailed reasoning about prior knowledge to make sense of the data. Such a hybrid strategy has been difficult to automate. Herein we consider crystal-structure phase mapping, a long standing challenge in materials science that is emblematic of the class of scientific problems whose automation constitutes a substantial advancement with respect to the grand challenge of high-throughput unsupervised scientific data interpretation.Crystal-structure phase mapping involves separating noisy mixtures of X-ray diffraction (XRD) patterns into the source XRD signals of the corresponding crystal structures, a task for

show abstract

Section: Competing Interestsmentioning

confidence: 99%

Automating Crystal-Structure Phase Mapping: Combining Deep Learning with Constraint Reasoning

Chen¹,

Bai²,

Ament³

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

“…Semi-supervised separation methods based on generative adversarial learning were proposed in [38], [39]. The key assumption of these methods is that estimated sources produced by an optimal separator should be indistinguishable from real sound sources, i.e., they should be samples drawn from the same distribution.…”

Section: Introductionmentioning

confidence: 99%

Finding Strength in Weakness: Learning to Separate Sounds with Weak Supervision

Pishdadian,

Wichern,

Roux

2019

Preprint

View full text Add to dashboard Cite

While there has been much recent progress using deep learning techniques to separate speech and music audio signals, these systems typically require large collections of isolated sources during the training process. When extending audio source separation algorithms to more general domains such as environmental monitoring, it may not be possible to obtain isolated signals for training. Here, we propose objective functions and network architectures that enable training a source separation system with weak labels. In this scenario, weak labels are defined in contrast with strong time-frequency (TF) labels such as those obtained from isolated sources, and refer either to frame-level weak labels where one only has access to the time periods when different sources are active in an audio mixture, or to clip-level weak labels that only indicate the presence or absence of sounds in an entire audio clip. We train a separator that estimates a TF mask for each type of sound event, using a sound event classifier as an assessor of the separator's performance to bridge the gap between the TF-level separation and the ground truth weak labels only available at the frame or clip level. Our objective function requires the classifier applied to a separated source to assign high probability to the class corresponding to that source and low probability to all other classes. The objective function also enforces that the separated sources sum up to the mixture. We benchmark the performance of our algorithm using synthetic mixtures of overlapping events created from a database of sounds recorded in urban environments. Compared to training a network using isolated sources, our model achieves somewhat lower but still significant SI-SDR improvement, even in scenarios with significant sound event overlap.

show abstract

Weakly Supervised Audio Source Separation via Spectrum Energy Preserved Wasserstein Learning

Cited by 2 publications

References 16 publications

Automating Crystal-Structure Phase Mapping: Combining Deep Learning with Constraint Reasoning

Automating Crystal-Structure Phase Mapping: Combining Deep Learning with Constraint Reasoning

Finding Strength in Weakness: Learning to Separate Sounds with Weak Supervision

Contact Info

Product

Resources

About