Variational Nested Dropout
Yufei Cui
Yushun Mao
Ziquan Liu
Qiao Li
Antoni Bert Chan
Tei-Wei Kuo
Chun Jason Xue
Nested dropout is a variant of dropout operation that is able to order network parameters or features based on the pre-defined importance du… (voir plus)ring training. It has been explored for: I. Constructing nested nets Cui et al. 2020, Cui et al. 2021: the nested nets are neural networks whose architectures can be adjusted instantly during testing time, e.g., based on computational constraints. The nested dropout implicitly ranks the network parameters, generating a set of sub-networks such that any smaller sub-network forms the basis of a larger one. II. Learning ordered representation Rippel et al. 2014: the nested dropout applied to the latent representation of a generative model (e.g., auto-encoder) ranks the features, enforcing explicit order of the dense representation over dimensions. However, the dropout rate is fixed as a hyper-parameter during the whole training process. For nested nets, when network parameters are removed, the performance decays in a human-specified trajectory rather than in a trajectory learned from data. For generative models, the importance of features is specified as a constant vector, restraining the flexibility of representation learning. To address the problem, we focus on the probabilistic counterpart of the nested dropout. We propose a variational nested dropout (VND) operation that draws samples of multi-dimensional ordered masks at a low cost, providing useful gradients to the parameters of nested dropout. Based on this approach, we design a Bayesian nested neural network that learns the order knowledge of the parameter distributions. We further exploit the VND under different generative models for learning ordered latent distributions. In experiments, we show that the proposed approach outperforms the nested network in terms of accuracy, calibration, and out-of-domain detection in classification tasks. It also outperforms the related generative models on data generation tasks.
Learning Domain Randomization Distributions for Training Robust Locomotion Policies
Melissa Mozian
Juan Higuera
This paper considers the problem of learning behaviors in simulation without knowledge of the precise dynamical properties of the target rob… (voir plus)ot platform(s). In this context, our learning goal is to mutually maximize task efficacy on each environment considered and generalization across the widest possible range of environmental conditions. The physical parameters of the simulator are modified by a component of our technique that learns the Domain Randomization (DR) that is appropriate at each learning epoch to maximally challenge the current behavior policy, without being overly challenging, which can hinder learning progress. This so-called sweet spot distribution is a selection of simulated domains with the following properties: 1) The trained policy should be successful in environments sampled from the domain randomization distribution; and 2) The DR distribution made as wide as possible, to increase variability in the environments. These properties aim to ensure the trajectories encountered in the target system are close to those observed during training, as existing methods in machine learning are better suited for interpolation than extrapolation. We show how adapting the DR distribution while training context-conditioned policies results in improvements on jump-start and asymptotic performance when transferring a learned policy to the target environment1.
Author response: Functional specialization within the inferior parietal lobes across cognitive domains
Ole Numssen
Gesa Hartwigsen
Correction to: The patient advisor, an organizational resource as a lever for an enhanced oncology patient experience (PAROLEonco): a longitudinal multiple case study protocol
Marie-Pascale Pomey
Michèle de Guise
Mado Desforges
Karine Bouchard
Cécile Vialaron
Louise Normandin
Monica Iliescu‐Nelea
Israël Fortin
Isabelle Ganache
Zeev Rosberger
Danielle Charpentier
L. Bélanger
Michel Dorval
Djahanchah Philip Ghadiri
Mélanie Lavoie-Tremblay
A. Boivin
Jean-François Pelletier
Nicolas Fernandez
Alain M. Danino
Deep learning identifies partially overlapping subnetworks in the human social brain
Hannah Kiesow
R. Nathan Spreng
Avram J. Holmes
Mallar Chakravarty
Andre Marquand
B.T. Thomas Yeo
Assessing the Impact: Does an Improvement to a Revenue Management System Lead to an Improved Revenue?
Greta Laage
Andrea Lodi
Learning with Gradient Descent and Weakly Convex Losses
Dominic Richards
We study the learning performance of gradient descent when the empirical risk is weakly convex, namely, the smallest negative eigenvalue of … (voir plus)the empirical risk's Hessian is bounded in magnitude. By showing that this eigenvalue can control the stability of gradient descent, generalisation error bounds are proven that hold under a wider range of step sizes compared to previous work. Out of sample guarantees are then achieved by decomposing the test error into generalisation, optimisation and approximation errors, each of which can be bounded and traded off with respect to algorithmic parameters, sample size and magnitude of this eigenvalue. In the case of a two layer neural network, we demonstrate that the empirical risk can satisfy a notion of local weak convexity, specifically, the Hessian's smallest eigenvalue during training can be controlled by the normalisation of the layers, i.e., network scaling. This allows test error guarantees to then be achieved when the population risk minimiser satisfies a complexity assumption. By trading off the network complexity and scaling, insights are gained into the implicit bias of neural network scaling, which are further supported by experimental findings.
Systematic detection of brain protein-coding genes under positive selection during primate evolution and their roles in cognition
Simon Malesys
Thomas Bourgeron
The human brain differs from that of other primates, but the genetic basis of these differences remains unclear. We investigated the evoluti… (voir plus)onary pressures acting on almost all human protein-coding genes (N = 11,667; 1:1 orthologs in primates) based on their divergence from those of early hominins, such as Neanderthals, and non-human primates. We confirm that genes encoding brain-related proteins are among the most strongly conserved protein-coding genes in the human genome. Combining our evolutionary pressure metrics for the protein-coding genome with recent data sets, we found that this conservation applied to genes functionally associated with the synapse and expressed in brain structures such as the prefrontal cortex and the cerebellum. Conversely, several genes presenting signatures commonly associated with positive selection appear as causing brain diseases or conditions, such as micro/macrocephaly, Joubert syndrome, dyslexia, and autism. Among those, a number of DNA damage response genes associated with microcephaly in humans such as BRCA1, NHEJ1, TOP3A, and RNF168 show strong signs of positive selection and might have played a role in human brain size expansion during primate evolution. We also showed that cerebellum granule neurons express a set of genes also presenting signatures of positive selection and that may have contributed to the emergence of fine motor skills and social cognition in humans. This resource is available online and can be used to estimate evolutionary constraints acting on a set of genes and to explore their relative contributions to human traits.
Adversarial score matching and improved sampling for image generation
Alexia Jolicoeur-Martineau
Rémi Piché-Taillefer
Remi Tachet des Combes
Denoising Score Matching with Annealed Langevin Sampling (DSM-ALS) has recently found success in generative modeling. The approach works by … (voir plus)first training a neural network to estimate the score of a distribution, and then using Langevin dynamics to sample from the data distribution assumed by the score network. Despite the convincing visual quality of samples, this method appears to perform worse than Generative Adversarial Networks (GANs) under the Fréchet Inception Distance, a standard metric for generative models. We show that this apparent gap vanishes when denoising the final Langevin samples using the score network. In addition, we propose two improvements to DSM-ALS: 1) Consistent Annealed Sampling as a more stable alternative to Annealed Langevin Sampling, and 2) a hybrid training formulation, composed of both Denoising Score Matching and adversarial objectives. By combining these two techniques and exploring different network architectures, we elevate score matching methods and obtain results competitive with state-of-the-art image generation on CIFAR-10.
CausalWorld: A Robotic Manipulation Benchmark for Causal Structure and Transfer Learning
Ossama Ahmed
Frederik Träuble
Anirudh Goyal
Alexander Neitz
Manuel Wüthrich
Bernhard Schölkopf
Stefan Bauer
Despite recent successes of reinforcement learning (RL), it remains a challenge for agents to transfer learned skills to related environment… (voir plus)s. To facilitate research addressing this problem, we proposeCausalWorld, a benchmark for causal structure and transfer learning in a robotic manipulation environment. The environment is a simulation of an open-source robotic platform, hence offering the possibility of sim-to-real transfer. Tasks consist of constructing 3D shapes from a set of blocks - inspired by how children learn to build complex structures. The key strength of CausalWorld is that it provides a combinatorial family of such tasks with common causal structure and underlying factors (including, e.g., robot and object masses, colors, sizes). The user (or the agent) may intervene on all causal variables, which allows for fine-grained control over how similar different tasks (or task distributions) are. One can thus easily define training and evaluation distributions of a desired difficulty level, targeting a specific form of generalization (e.g., only changes in appearance or object mass). Further, this common parametrization facilitates defining curricula by interpolating between an initial and a target task. While users may define their own task distributions, we present eight meaningful distributions as concrete benchmarks, ranging from simple to very challenging, all of which require long-horizon planning as well as precise low-level motor control. Finally, we provide baseline results for a subset of these tasks on distinct training curricula and corresponding evaluation protocols, verifying the feasibility of the tasks in this benchmark.
Integrating Categorical Semantics into Unsupervised Domain Translation
Samuel Lavoie-Marchildon
Faruk Ahmed
While unsupervised domain translation (UDT) has seen a lot of success recently, we argue that mediating its translation via categorical sema… (voir plus)ntic features could broaden its applicability. In particular, we demonstrate that categorical semantics improves the translation between perceptually different domains sharing multiple object categories. We propose a method to learn, in an unsupervised manner, categorical semantic features (such as object labels) that are invariant of the source and target domains. We show that conditioning the style encoder of unsupervised domain translation methods on the learned categorical semantics leads to a translation preserving the digits on MNIST
Iterated learning for emergent systematicity in VQA
Ankit Vani
Max Schwarzer
Yuchen Lu
Eeshan Dhekane
Although neural module networks have an architectural bias towards compositionality, they require gold standard layouts to generalize system… (voir plus)atically in practice. When instead learning layouts and modules jointly, compositionality does not arise automatically and an explicit pressure is necessary for the emergence of layouts exhibiting the right structure. We propose to address this problem using iterated learning, a cognitive science theory of the emergence of compositional languages in nature that has primarily been applied to simple referential games in machine learning. Considering the layouts of module networks as samples from an emergent language, we use iterated learning to encourage the development of structure within this language. We show that the resulting layouts support systematic generalization in neural agents solving the more complex task of visual question-answering. Our regularized iterated learning method can outperform baselines without iterated learning on SHAPES-SyGeT (SHAPES Systematic Generalization Test), a new split of the SHAPES dataset we introduce to evaluate systematic generalization, and on CLOSURE, an extension of CLEVR also designed to test systematic generalization. We demonstrate superior performance in recovering ground-truth compositional program structure with limited supervision on both SHAPES-SyGeT and CLEVR.