Improving Source Separation by Explicitly Modeling Dependencies between Sources
Ethan Manilow
Curtis Hawthorne
Bryan Pardo
Jesse Engel
We propose a new method for training a supervised source separation system that aims to learn the interdependent relationships between all c… (see more)ombinations of sources in a mixture. Rather than independently estimating each source from a mix, we reframe the source separation problem as an Orderless Neural Autoregressive Density Estimator (NADE), and estimate each source from both the mix and a random subset of the other sources. We adapt a standard source separation architecture, Demucs, with additional inputs for each individual source, in addition to the input mixture. We randomly mask these input sources during training so that the network learns the conditional dependencies between the sources. By pairing this training method with a blocked Gibbs sampling procedure at inference time, we demonstrate that the network can iteratively improve its separation performance by conditioning a source estimate on its earlier source estimates. Experiments on two source separation datasets show that training a Demucs model with an Orderless NADE approach and using Gibbs sampling (up to 512 steps) at inference time strongly outperforms a Demucs baseline that uses a standard regression loss and direct (one step) estimation of sources.
Real-M: Towards Speech Separation on Real Mixtures
Samuele Cornell
François Grondin
In recent years, deep learning based source separation has achieved impressive results. Most studies, however, still evaluate separation mod… (see more)els on synthetic datasets, while the performance of state-of-the-art techniques on in-the-wild speech data remains an open question. This paper contributes to fill this gap in two ways. First, we release the REAL-M dataset, a crowd-sourced corpus of real-life mixtures. Secondly, we address the problem of performance evaluation of real-life mixtures, where the ground truth is not available. We bypass this issue by carefully designing a blind Scale-Invariant Signal-to-Noise Ratio (SI-SNR) neural estimator. Through a user study, we show that our estimator reliably evaluates the separation performance on real mixtures, i.e. we observe that the performance predictions of the SI-SNR estimator correlate well with human opinions. Moreover, when evaluating popular speech separation models, we observe that the performance trends predicted by our estimator on the REAL-M dataset closely follow the performance trends achieved on synthetic benchmarks.
A Remedy For Distributional Shifts Through Expected Domain Translation
Jean-Christophe Gagnon-Audet
Soroosh Shahtalebi
Frank Rudzicz
Machine learning models often fail to generalize to unseen domains due to the distributional shifts. A family of such shifts, “correlation… (see more) shifts,” is caused by spurious correlations in the data. It is studied under the overarching topic of “domain generalization.” In this work, we employ multi-modal translation networks to tackle the correlation shifts that appear when data is sampled out-of-distribution. Learning a generative model from training domains enables us to translate each training sample under the special characteristics of other possible domains. We show that by training a predictor solely on the generated samples, the spurious correlations in training domains average out, and the invariant features corresponding to true correlations emerge. Our proposed technique, Expected Domain Translation (EDT), is benchmarked on the Colored MNIST dataset and drastically improves the state-of-the-art classification accuracy by 38% with train-domain validation model selection.
Roboethics as a Design Challenge: Lessons Learned from the Roboethics to Design and Development Competition
Jimin Rhim
Cheng Lin
Alexander Werner
Brandon DeHart
Vivian Qiang
How do we make concrete progress towards de-signing robots that can navigate ethically sensitive contexts? Almost two decades after the word… (see more) ‘roboethics’ was coined, translating interdisciplinary roboethics discussions into techni-cal design still remains a daunting task. This paper describes our first attempt at addressing these challenges through a roboethics-themed design competition. The design competition setting allowed us to (a) formulate ethical considerations as an engineering design task that anyone with basic programming skills can tackle; and (b) develop a prototype evaluation scheme that incorporates diverse normative perspectives of multiple stakeholders. The initial implementation of the competition was held online at the RO-MAN 2021 conference. The competition task involved programming a simulated mobile robot (TIAGo) that delivers items for individuals in the home environment, where many of these tasks involve ethically sensitive con-texts (e.g., an underage family member asks for an alcoholic drink). This paper outlines our experiences implementing the competition and the lessons we learned. We highlight design competitions as a promising mechanism to enable a new wave of roboethics research equipped with technical design solutions.
Tyger: Task-Type-Generic Active Learning for Molecular Property Prediction
Kuangqi Zhou
Kaixin Wang
Jiashi Feng
Tingyang Xu
Xinchao Wang
How to accurately predict the properties of molecules is an essential problem in AI-driven drug discovery, which generally requires a large … (see more)amount of annotation for training deep learning models. Annotating molecules, however, is quite costly because it requires lab experiments conducted by experts. To reduce annotation cost, deep Active Learning (AL) methods are developed to select only the most representative and informative data for annotating. However, existing best deep AL methods are mostly developed for a single type of learning task (e.g., single-label classification), and hence may not perform well in molecular property prediction that involves various task types. In this paper, we propose a Task-type-generic active learning framework (termed Tyger) that is able to handle different types of learning tasks in a unified manner. The key is to learn a chemically-meaningful embedding space and perform active selection fully based on the embeddings, instead of relying on task-type-specific heuristics (e.g., class-wise prediction probability) as done in existing works. Specifically, for learning the embedding space, we instantiate a querying module that learns to translate molecule graphs into corresponding SMILES strings. Furthermore, to ensure that samples selected from the space are both representative and informative, we propose to shape the embedding space by two learning objectives, one based on domain knowledge and the other leveraging feedback from the task learner (i.e., model that performs the learning task at hand). We conduct extensive experiments on benchmark datasets of different task types. Experimental results show that Tyger consistently achieves high AL performance on molecular property prediction, outperforming baselines by a large margin. We also perform ablative experiments to verify the effectiveness of each component in Tyger.
Better Modeling the Programming World with Code Concept Graphs-augmented Multi-modal Learning
Martin Weyssow
Houari Sahraoui
The progress made in code modeling has been tremendous in recent years thanks to the design of natural language processing learning approach… (see more)es based on state-of-the-art model architectures. Nevertheless, we believe that the current state-of-the-art does not focus enough on the full potential that data may bring to a learning process in software engineering. Our vision articulates on the idea of leveraging multi-modal learning approaches to modeling the programming world. In this paper, we investigate one of the underlying idea of our vision whose objective based on concept graphs of identifiers aims at leveraging high-level relationships between domain concepts manipulated through particular language constructs. In particular, we propose to enhance an existing pretrained language model of code by joint-learning it with a graph neural network based on our concept graphs. We conducted a preliminary evaluation that shows gain of effectiveness of the models for code search using a simple joint-learning method and prompts us to further investigate our research vision.
Coordinating Policies Among Multiple Agents via an Intelligent Communication Channel
Dianbo Liu
Cristian Meo
Anirudh Goyal
Tianmin Shu
Michael Curtis Mozer
Nicolas Heess
In Multi-Agent Reinforcement Learning (MARL), specialized channels are often introduced that allow agents to communicate directly with one a… (see more)nother. In this paper, we propose an alternative approach whereby agents communicate through an intelligent facilitator that learns to sift through and interpret signals provided by all agents to improve the agents’ collective performance. To ensure that this facilitator does not become a centralized controller, agents are incentivized to reduce their dependence on the messages it conveys, and the messages can only influence the selection of a policy from a fixed set, not instantaneous actions given the policy. We demonstrate the strength of this architecture over existing baselines on several cooperative MARL environments.
Bayesian Structure Learning with Generative Flow Networks
In Bayesian structure learning, we are interested in inferring a distribution over the directed acyclic graph (DAG) structure of Bayesian ne… (see more)tworks, from data. Defining such a distribution is very challenging, due to the combinatorially large sample space, and approximations based on MCMC are often required. Recently, a novel class of probabilistic models, called Generative Flow Networks (GFlowNets), have been introduced as a general framework for generative modeling of discrete and composite objects, such as graphs. In this work, we propose to use a GFlowNet as an alternative to MCMC for approximating the posterior distribution over the structure of Bayesian networks, given a dataset of observations. Generating a sample DAG from this approximate distribution is viewed as a sequential decision problem, where the graph is constructed one edge at a time, based on learned transition probabilities. Through evaluation on both simulated and real data, we show that our approach, called DAG-GFlowNet, provides an accurate approximation of the posterior over DAGs, and it compares favorably against other methods based on MCMC or variational inference.
Hardware Architecture for Guessing Random Additive Noise Decoding Markov Order (GRAND-MO)
Syed Mohsin Abbas
Marwan Jalaleddine
Pattern learning reveals brain asymmetry to be linked to socioeconomic status
Timm B Poeppl
Emile Dimas
Katrin Sakreida
Julius M Kernbach
Ross D Markello
Oliver Schöffski
Alain Dagher
Philipp Koellinger
Gideon Nave
Martha J Farah
Bratislav Mišić
Abstract Socioeconomic status (SES) anchors individuals in their social network layers. Our embedding in the societal fabric resonates with … (see more)habitus, world view, opportunity, and health disparity. It remains obscure how distinct facets of SES are reflected in the architecture of the central nervous system. Here, we capitalized on multivariate multi-output learning algorithms to explore possible imprints of SES in gray and white matter structure in the wider population (n ≈ 10,000 UK Biobank participants). Individuals with higher SES, compared with those with lower SES, showed a pattern of increased region volumes in the left brain and decreased region volumes in the right brain. The analogous lateralization pattern emerged for the fiber structure of anatomical white matter tracts. Our multimodal findings suggest hemispheric asymmetry as an SES-related brain signature, which was consistent across six different indicators of SES: degree, education, income, job, neighborhood and vehicle count. Hence, hemispheric specialization may have evolved in human primates in a way that reveals crucial links to SES.
Privacy-Aware Compression for Federated Data Analysis
Kamalika Chaudhuri
Chuan Guo
Federated data analytics is a framework for distributed data analysis where a server compiles noisy responses from a group of distributed lo… (see more)w-bandwidth user devices to estimate aggregate statistics. Two major challenges in this framework are privacy, since user data is often sensitive, and compression, since the user devices have low network bandwidth. Prior work has addressed these challenges separately by combining standard compression algorithms with known privacy mechanisms. In this work, we take a holistic look at the problem and design a family of privacy-aware compression mechanisms that work for any given communication budget. We first propose a mechanism for transmitting a single real number that has optimal variance under certain conditions. We then show how to extend it to metric differential privacy for location privacy use-cases, as well as vectors, for application to federated learning. Our experiments illustrate that our mechanism can lead to better utility vs. compression trade-offs for the same privacy loss in a number of settings.
Temporal Abstractions-Augmented Temporally Contrastive Learning: An Alternative to the Laplacian in RL
Akram Erraqabi
Marlos C. Machado
Mingde Zhao
Sainbayar Sukhbaatar
Alessandro Lazaric
Ludovic Denoyer
In reinforcement learning, the graph Laplacian has proved to be a valuable tool in the task-agnostic setting, with applications ranging from… (see more) skill discovery to reward shaping. Recently, learning the Laplacian representation has been framed as the optimization of a temporally-contrastive objective to overcome its computational limitations in large (or continuous) state spaces. However, this approach requires uniform access to all states in the state space, overlooking the exploration problem that emerges during the representation learning process. In this work, we propose an alternative method that is able to recover, in a non-uniform-prior setting, the expressiveness and the desired properties of the Laplacian representation. We do so by combining the representation learning with a skill-based covering policy, which provides a better training distribution to extend and refine the representation. We also show that a simple augmentation of the representation objective with the learned temporal abstractions improves dynamics-awareness and helps exploration. We find that our method succeeds as an alternative to the Laplacian in the non-uniform setting and scales to challenging continuous control environments. Finally, even if our method is not optimized for skill discovery, the learned skills can successfully solve difficult continuous navigation tasks with sparse rewards, where standard skill discovery approaches are no so effective.