Paper Group NANR 17
Logic and the 2-Simplicial Transformer. Variational Diffusion Autoencoders with Random Walk Sampling. Manifold Learning and Alignment with Generative Adversarial Networks. Poisoning Attacks with Generative Adversarial Nets. Locality and Compositionality in Zero-Shot Learning. Off-Policy Actor-Critic with Shared Experience Replay. Unsupervised Learn …
Logic and the 2-Simplicial Transformer
| Title | Logic and the 2-Simplicial Transformer |
| Authors | Anonymous |
| Abstract | We introduce the 2-simplicial Transformer, an extension of the Transformer which includes a form of higher-dimensional attention generalising the dot-product attention, and uses this attention to update entity representations with tensor products of value vectors. We show that this architecture is a useful inductive bias for logical reasoning in the context of deep reinforcement learning. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rkecJ6VFvr |
| https://openreview.net/pdf?id=rkecJ6VFvr | |
| PWC | https://paperswithcode.com/paper/logic-and-the-2-simplicial-transformer-1 |
| Repo | |
| Framework | |
Variational Diffusion Autoencoders with Random Walk Sampling
| Title | Variational Diffusion Autoencoders with Random Walk Sampling |
| Authors | Anonymous |
| Abstract | Variational inference (VI) methods and especially variational autoencoders (VAEs) specify scalable generative models that enjoy an intuitive connection to manifold learning — with many default priors the posterior/likelihood pair $q(zx)$/$p(xz)$ can be viewed as an approximate homeomorphism (and its inverse) between the data manifold and a latent Euclidean space. However, these approximations are well-documented to become degenerate in training. Unless the subjective prior is carefully chosen, the topologies of the prior and data distributions often will not match. Conversely, diffusion maps (DM) automatically \textit{infer} the data topology and enjoy a rigorous connection to manifold learning, but do not scale easily or provide the inverse homeomorphism. In this paper, we propose \textbf{a)} a principled measure for recognizing the mismatch between data and latent distributions and \textbf{b)} a method that combines the advantages of variational inference and diffusion maps to learn a homeomorphic generative model. The measure, the \textit{locally bi-Lipschitz property}, is a sufficient condition for a homeomorphism and easy to compute and interpret. The method, the \textit{variational diffusion autoencoder} (VDAE), is a novel generative algorithm that first infers the topology of the data distribution, then models a diffusion random walk over the data. To achieve efficient computation in VDAEs, we use stochastic versions of both variational inference and manifold learning optimization. We prove approximation theoretic results for the dimension dependence of VDAEs, and that locally isotropic sampling in the latent space results in a random walk over the reconstructed manifold. Finally, we demonstrate the utility of our method on various real and synthetic datasets, and show that it exhibits performance superior to other generative models. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rkg8FJBYDS |
| https://openreview.net/pdf?id=rkg8FJBYDS | |
| PWC | https://paperswithcode.com/paper/variational-diffusion-autoencoders-with |
| Repo | |
| Framework | |
Manifold Learning and Alignment with Generative Adversarial Networks
| Title | Manifold Learning and Alignment with Generative Adversarial Networks |
| Authors | Anonymous |
| Abstract | We present a generative adversarial network (GAN) that conducts manifold learning and alignment (MLA): A task to learn the multi-manifold structure underlying data and to align those manifolds without any correspondence information. Our main idea is to exploit the powerful abstraction ability of encoder architecture. Specifically, we define multiple generators to model multiple manifolds, but in a particular way that their inverse maps can be commonly represented by a single smooth encoder. Then, the abstraction ability of the encoder enforces semantic similarities between the generators and gives a plausibly aligned embedding in the latent space. In experiments with MNIST, 3D-Chair, and UT-Zap50k datasets, we demonstrate the superiority of our model in learning the manifolds by FID scores and in aligning the manifolds by disentanglement scores. Furthermore, by virtue of the abstractive modeling, we show that our model can generate data from an untrained manifold, which is unique to our model. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=r1eCukHYDH |
| https://openreview.net/pdf?id=r1eCukHYDH | |
| PWC | https://paperswithcode.com/paper/manifold-learning-and-alignment-with |
| Repo | |
| Framework | |
Poisoning Attacks with Generative Adversarial Nets
| Title | Poisoning Attacks with Generative Adversarial Nets |
| Authors | Anonymous |
| Abstract | Machine learning algorithms are vulnerable to poisoning attacks: An adversary can inject malicious points in the training dataset to influence the learning process and degrade the algorithm’s performance. Optimal poisoning attacks have already been proposed to evaluate worst-case scenarios, modelling attacks as a bi-level optimization problem. Solving these problems is computationally demanding and has limited applicability for some models such as deep networks. In this paper we introduce a novel generative model to craft systematic poisoning attacks against machine learning classifiers generating adversarial training examples, i.e. samples that look like genuine data points but that degrade the classifier’s accuracy when used for training. We propose a Generative Adversarial Net with three components: generator, discriminator, and the target classifier. This approach allows us to model naturally the detectability constrains that can be expected in realistic attacks and to identify the regions of the underlying data distribution that can be more vulnerable to data poisoning. Our experimental evaluation shows the effectiveness of our attack to compromise machine learning classifiers, including deep networks. |
| Tasks | data poisoning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Bke-6pVKvB |
| https://openreview.net/pdf?id=Bke-6pVKvB | |
| PWC | https://paperswithcode.com/paper/poisoning-attacks-with-generative-adversarial-1 |
| Repo | |
| Framework | |
Locality and Compositionality in Zero-Shot Learning
| Title | Locality and Compositionality in Zero-Shot Learning |
| Authors | Anonymous |
| Abstract | In this work we study locality and compositionality in the context of learning representations for Zero Shot Learning (ZSL). In order to well-isolate the importance of these properties in learned representations, we impose the additional constraint that, differently from most recent work in ZSL, no pre-training on different datasets (e.g. ImageNet) is performed. The results of our experiment show how locality, in terms of small parts of the input, and compositionality, i.e. how well can the learned representations be expressed as a function of a smaller vocabulary, are both deeply related to generalization and motivate the focus on more local-aware models in future research directions for representation learning. |
| Tasks | Representation Learning, Zero-Shot Learning |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Hye_V0NKwr |
| https://openreview.net/pdf?id=Hye_V0NKwr | |
| PWC | https://paperswithcode.com/paper/locality-and-compositionality-in-zero-shot |
| Repo | |
| Framework | |
Off-Policy Actor-Critic with Shared Experience Replay
| Title | Off-Policy Actor-Critic with Shared Experience Replay |
| Authors | Anonymous |
| Abstract | We investigate the combination of actor-critic reinforcement learning algorithms with uniform large-scale experience replay and propose solutions for two challenges: (a) efficient actor-critic learning with experience replay (b) stability of very off-policy learning. We employ those insights to accelerate hyper-parameter sweeps in which all participating agents run concurrently and share their experience via a common replay module. To this end we analyze the bias-variance tradeoffs in V-trace, a form of importance sampling for actor-critic methods. Based on our analysis, we then argue for mixing experience sampled from replay with on-policy experience, and propose a new trust region scheme that scales effectively to data distributions where V-trace becomes unstable. We provide extensive empirical validation of the proposed solution. We further show the benefits of this setup by demonstrating state-of-the-art data efficiency on Atari among agents trained up until 200M environment frames. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=HygaikBKvS |
| https://openreview.net/pdf?id=HygaikBKvS | |
| PWC | https://paperswithcode.com/paper/off-policy-actor-critic-with-shared |
| Repo | |
| Framework | |
Unsupervised Learning of Node Embeddings by Detecting Communities
| Title | Unsupervised Learning of Node Embeddings by Detecting Communities |
| Authors | Anonymous |
| Abstract | We present Deep MinCut (DMC), an unsupervised approach to learn node embeddings for graph-structured data. It derives node representations based on their membership in communities. As such, the embeddings directly provide interesting insights into the graph structure, so that the separate node clustering step of existing methods is no longer needed. DMC learns both, node embeddings and communities, simultaneously by minimizing the mincut loss, which captures the number of connections between communities. Striving for high scalability, we also propose a training process for DMC based on minibatches. We provide empirical evidence that the communities learned by DMC are meaningful and that the node embeddings are competitive in different node classification benchmarks. |
| Tasks | Node Classification |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Byl3K2VtwB |
| https://openreview.net/pdf?id=Byl3K2VtwB | |
| PWC | https://paperswithcode.com/paper/unsupervised-learning-of-node-embeddings-by |
| Repo | |
| Framework | |
Neural Maximum Common Subgraph Detection with Guided Subgraph Extraction
| Title | Neural Maximum Common Subgraph Detection with Guided Subgraph Extraction |
| Authors | Anonymous |
| Abstract | Maximum Common Subgraph (MCS) is defined as the largest subgraph that is commonly present in both graphs of a graph pair. Exact MCS detection is NP-hard, and its state-of-the-art exact solver based on heuristic search is slow in practice without any time complexity guarantee. Given the huge importance of this task yet the lack of fast solver, we propose an efficient MCS detection algorithm, NeuralMCS, consisting of a novel neural network model that learns the node-node correspondence from the ground-truth MCS result, and a subgraph extraction procedure that uses the neural network output as guidance for final MCS prediction. The whole model guarantees polynomial time complexity with respect to the number of the nodes of the larger of the two input graphs. Experiments on four real graph datasets show that the proposed model is 48.1x faster than the exact solver and more accurate than all the existing competitive approximate approaches to MCS detection. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BJgcwh4FwS |
| https://openreview.net/pdf?id=BJgcwh4FwS | |
| PWC | https://paperswithcode.com/paper/neural-maximum-common-subgraph-detection-with |
| Repo | |
| Framework | |
On the interaction between supervision and self-play in emergent communication
| Title | On the interaction between supervision and self-play in emergent communication |
| Authors | Anonymous |
| Abstract | A promising approach for teaching artificial agents to use natural language involves using human-in-the-loop training. However, recent work suggests that current machine learning methods are too data inefficient to be trained in this way from scratch. In this paper, we investigate the relationship between two categories of learning signals with the ultimate goal of improving sample efficiency: imitating human language data via supervised learning, and maximizing reward in a simulated multi-agent environment via self-play (as done in emergent communication), and introduce the term \textit{supervised self-play (S2P)} for algorithms using both of these signals. We find that first training agents via supervised learning on human data followed by self-play outperforms the converse, suggesting that it is not beneficial to emerge languages from scratch. We then empirically investigate various S2P schedules that begin with supervised learning in two environments: a Lewis signaling game with symbolic inputs, and an image-based referential game with natural language descriptions. Lastly, we introduce population based approaches to S2P, which further improves the performance over single-agent methods. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rJxGLlBtwH |
| https://openreview.net/pdf?id=rJxGLlBtwH | |
| PWC | https://paperswithcode.com/paper/on-the-interaction-between-supervision-and |
| Repo | |
| Framework | |
Reducing Computation in Recurrent Networks by Selectively Updating State Neurons
| Title | Reducing Computation in Recurrent Networks by Selectively Updating State Neurons |
| Authors | Anonymous |
| Abstract | Recurrent Neural Networks (RNN) are the state-of-the-art approach to sequential learning. However, standard RNNs use the same amount of computation at each timestep, regardless of the input data. As a result, even for high-dimensional hidden states, all dimensions are updated at each timestep regardless of the recurrent memory cell. Reducing this rigid assumption could allow for models with large hidden states to perform inference more quickly. Intuitively, not all hidden state dimensions need to be recomputed from scratch at each timestep. Thus, recent methods have begun studying this problem by imposing mainly a priori-determined patterns for updating the state. In contrast, we now design a fully-learned approach, SA-RNN, that augments any RNN by predicting discrete update patterns at the fine granularity of independent hidden state dimensions through the parameterization of a distribution of update-likelihoods driven entirely by the input data. We achieve this without imposing assumptions on the structure of the update pattern. Better yet, our method adapts the update patterns online, allowing different dimensions to be updated conditional to the input. To learn which to update, the model solves a multi-objective optimization problem, maximizing accuracy while minimizing the number of updates based on a unified control. Using publicly-available datasets we demonstrate that our method consistently achieves higher accuracy with fewer updates compared to state-of-the-art alternatives. Additionally, our method can be directly applied to a wide variety of models containing RNN architectures. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SkeP3yBFDS |
| https://openreview.net/pdf?id=SkeP3yBFDS | |
| PWC | https://paperswithcode.com/paper/reducing-computation-in-recurrent-networks-by |
| Repo | |
| Framework | |
Dual-module Inference for Efficient Recurrent Neural Networks
| Title | Dual-module Inference for Efficient Recurrent Neural Networks |
| Authors | Anonymous |
| Abstract | Using Recurrent Neural Networks (RNNs) in sequence modeling tasks is promising in delivering high-quality results but challenging to meet stringent latency requirements because of the memory-bound execution pattern of RNNs. We propose a big-little dual-module inference to dynamically skip unnecessary memory access and computation to speedup RNN inference. Leveraging the error-resilient feature of nonlinear activation functions used in RNNs, we propose to use a lightweight little module that approximates the original RNN layer, which is referred to as the big module, to compute activations of the insensitive region that are more error-resilient. The expensive memory access and computation of the big module can be reduced as the results are only used in the sensitive region. Our method can reduce the overall memory access by 40% on average and achieve 1.54x to 1.75x speedup on CPU-based server platform with negligible impact on model quality. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=SJe3KCNKPr |
| https://openreview.net/pdf?id=SJe3KCNKPr | |
| PWC | https://paperswithcode.com/paper/dual-module-inference-for-efficient-recurrent |
| Repo | |
| Framework | |
Towards Physics-informed Deep Learning for Turbulent Flow Prediction
| Title | Towards Physics-informed Deep Learning for Turbulent Flow Prediction |
| Authors | Anonymous |
| Abstract | While deep learning has shown tremendous success in a wide range of domains, it remains a grand challenge to incorporate physical principles in a systematic manner to the design, training, and inference of such models. In this paper, we aim to predict turbulent flow by learning its highly nonlinear dynamics from spatiotemporal velocity fields of large-scale fluid flow simulations of relevance to turbulence modeling and climate modeling. We adopt a hybrid approach by marrying two well-established turbulent flow simulation techniques with deep learning. Specifically, we introduce trainable spectral filters in a coupled model of Reynolds-averaged Navier-Stokes (RANS) and Large Eddy Simulation (LES), followed by a specialized U-net for prediction. Our approach, which we call Turbulent-Flow Net (TF-Net), is grounded in a principled physics model, yet offers the flexibility of learned representations. We compare our model, TF-Net, with state-of-the-art baselines and observe significant reductions in error for predictions 60 frames ahead. Most significantly, our method predicts physical fields that obey desirable physical characteristics, such as conservation of mass, whilst faithfully emulating the turbulent kinetic energy field and spectrum, which are critical for accurate prediction of turbulent flows. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=Hkg5lAEtvS |
| https://openreview.net/pdf?id=Hkg5lAEtvS | |
| PWC | https://paperswithcode.com/paper/towards-physics-informed-deep-learning-for |
| Repo | |
| Framework | |
Difference-Seeking Generative Adversarial Network–Unseen Sample Generation
| Title | Difference-Seeking Generative Adversarial Network–Unseen Sample Generation |
| Authors | Anonymous |
| Abstract | Unseen data, which are not samples from the distribution of training data and are difficult to collect, have exhibited importance in numerous applications, ({\em e.g.,} novelty detection, semi-supervised learning, and adversarial training). In this paper, we introduce a general framework called \textbf{d}ifference-\textbf{s}eeking \textbf{g}enerative \textbf{a}dversarial \textbf{n}etwork (DSGAN), to generate various types of unseen data. Its novelty is the consideration of the probability density of the unseen data distribution as the difference between two distributions $p_{\bar{d}}$ and $p_{d}$ whose samples are relatively easy to collect. The DSGAN can learn the target distribution, $p_{t}$, (or the unseen data distribution) from only the samples from the two distributions, $p_{d}$ and $p_{\bar{d}}$. In our scenario, $p_d$ is the distribution of the seen data, and $p_{\bar{d}}$ can be obtained from $p_{d}$ via simple operations, so that we only need the samples of $p_{d}$ during the training. Two key applications, semi-supervised learning and novelty detection, are taken as case studies to illustrate that the DSGAN enables the production of various unseen data. We also provide theoretical analyses about the convergence of the DSGAN. |
| Tasks | |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=rygjmpVFvB |
| https://openreview.net/pdf?id=rygjmpVFvB | |
| PWC | https://paperswithcode.com/paper/difference-seeking-generative-adversarial-1 |
| Repo | |
| Framework | |
Actor-Critic Approach for Temporal Predictive Clustering
| Title | Actor-Critic Approach for Temporal Predictive Clustering |
| Authors | Anonymous |
| Abstract | Due to the wider availability of modern electronic health records (EHR), patient care data is often being stored in the form of time-series. Clustering such time-series data is crucial for patient phenotyping, anticipating patients’ prognoses by identifying “similar” patients, and designing treatment guidelines that are tailored to homogeneous patient subgroups. In this paper, we develop a deep learning approach for clustering time-series data, where each cluster comprises patients who share similar future outcomes of interest (e.g., adverse events, the onset of comorbidities, etc.). The clustering is carried out by using our novel loss functions that encourage each cluster to have homogeneous future outcomes. We adopt actor-critic models to allow “back-propagation” through the sampling process that is required for assigning clusters to time-series inputs. Experiments on two real-world datasets show that our model achieves superior clustering performance over state-of-the-art benchmarks and identifies meaningful clusters that can be translated into actionable information for clinical decision-making. |
| Tasks | Decision Making, Time Series, Time Series Clustering |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=r1ln504YvH |
| https://openreview.net/pdf?id=r1ln504YvH | |
| PWC | https://paperswithcode.com/paper/actor-critic-approach-for-temporal-predictive |
| Repo | |
| Framework | |
Clustered Reinforcement Learning
| Title | Clustered Reinforcement Learning |
| Authors | Anonymous |
| Abstract | Exploration strategy design is one of the challenging problems in reinforcement learning~(RL), especially when the environment contains a large state space or sparse rewards. During exploration, the agent tries to discover novel areas or high reward~(quality) areas. In most existing methods, the novelty and quality in the neighboring area of the current state are not well utilized to guide the exploration of the agent. To tackle this problem, we propose a novel RL framework, called \underline{c}lustered \underline{r}einforcement \underline{l}earning~(CRL), for efficient exploration in RL. CRL adopts clustering to divide the collected states into several clusters, based on which a bonus reward reflecting both novelty and quality in the neighboring area~(cluster) of the current state is given to the agent. Experiments on several continuous control tasks and several Atari-2600 games show that CRL can outperform other state-of-the-art methods to achieve the best performance in most cases. |
| Tasks | Atari Games, Continuous Control, Efficient Exploration |
| Published | 2020-01-01 |
| URL | https://openreview.net/forum?id=BylD9eSYPS |
| https://openreview.net/pdf?id=BylD9eSYPS | |
| PWC | https://paperswithcode.com/paper/clustered-reinforcement-learning-1 |
| Repo | |
| Framework | |
