February 1, 2020

3393 words 16 mins read

Paper Group AWR 293

PaccMann$^{RL}$: Designing anticancer drugs from transcriptomic data via reinforcement learning. Finding Archetypal Spaces Using Neural Networks. GAN-based Virtual Re-Staining: A Promising Solution for Whole Slide Image Analysis. Audio Denoising with Deep Network Priors. Systematic quantitative analyses reveal the folk-zoological knowledge embedded …

PaccMann$^{RL}$: Designing anticancer drugs from transcriptomic data via reinforcement learning


Title	PaccMann$^{RL}$: Designing anticancer drugs from transcriptomic data via reinforcement learning
Authors	Jannis Born, Matteo Manica, Ali Oskooei, Joris Cadow, María Rodríguez Martínez
Abstract	With the advent of deep generative models in computational chemistry, in silico anticancer drug design has undergone an unprecedented transformation. While state-of-the-art deep learning approaches have shown potential in generating compounds with desired chemical properties, they disregard the genetic profile and properties of the target disease. Here, we introduce the first generative model capable of tailoring anticancer compounds for a specific biomolecular profile. Using a RL framework, the transcriptomic profiles of cancer cells are used as a context for the generation of candidate molecules. Our molecule generator combines two separately pretrained variational autoencoders (VAEs) - the first VAE encodes transcriptomic profiles into a smooth, latent space which in turn is used to condition a second VAE to generate novel molecular structures on the given transcriptomic profile. The generative process is optimized through PaccMann, a previously developed drug sensitivity prediction model to obtain effective anticancer compounds for the given context (i.e., transcriptomic profile). We demonstrate how the molecule generation can be biased towards compounds with high predicted inhibitory effect against individual cell lines or specific cancer sites. We verify our approach by investigating candidate drugs generated against specific cancer types and find the highest structural similarity to existing compounds with known efficacy against these cancer types. We envision our approach to transform in silico anticancer drug design by leveraging the biomolecular characteristics of the disease in order to increase success rates in lead compound discovery.
Tasks
Published	2019-08-29
URL	https://arxiv.org/abs/1909.05114v3
PDF	https://arxiv.org/pdf/1909.05114v3.pdf
PWC	https://paperswithcode.com/paper/reinforcement-learning-driven-de-novo-design
Repo	https://github.com/PaccMann/paccmann_rl
Framework	none

Finding Archetypal Spaces Using Neural Networks


Title	Finding Archetypal Spaces Using Neural Networks
Authors	David van Dijk, Daniel Burkhardt, Matthew Amodio, Alex Tong, Guy Wolf, Smita Krishnaswamy
Abstract	Archetypal analysis is a data decomposition method that describes each observation in a dataset as a convex combination of “pure types” or archetypes. These archetypes represent extrema of a data space in which there is a trade-off between features, such as in biology where different combinations of traits provide optimal fitness for different environments. Existing methods for archetypal analysis work well when a linear relationship exists between the feature space and the archetypal space. However, such methods are not applicable to systems where the feature space is generated non-linearly from the combination of archetypes, such as in biological systems or image transformations. Here, we propose a reformulation of the problem such that the goal is to learn a non-linear transformation of the data into a latent archetypal space. To solve this problem, we introduce Archetypal Analysis network (AAnet), which is a deep neural network framework for learning and generating from a latent archetypal representation of data. We demonstrate state-of-the-art recovery of ground-truth archetypes in non-linear data domains, show AAnet can generate from data geometry rather than from data density, and use AAnet to identify biologically meaningful archetypes in single-cell gene expression data.
Tasks
Published	2019-01-25
URL	https://arxiv.org/abs/1901.09078v2
PDF	https://arxiv.org/pdf/1901.09078v2.pdf
PWC	https://paperswithcode.com/paper/finding-archetypal-spaces-for-data-using
Repo	https://github.com/KrishnaswamyLab/AAnet
Framework	tf

GAN-based Virtual Re-Staining: A Promising Solution for Whole Slide Image Analysis


Title	GAN-based Virtual Re-Staining: A Promising Solution for Whole Slide Image Analysis
Authors	Zhaoyang Xu, Carlos Fernández Moro, Béla Bozóky, Qianni Zhang
Abstract	Histopathological cancer diagnosis is based on visual examination of stained tissue slides. Hematoxylin and eosin (H&E) is a standard stain routinely employed worldwide. It is easy to acquire and cost effective, but cells and tissue components show low-contrast with varying tones of dark blue and pink, which makes difficult visual assessments, digital image analysis, and quantifications. These limitations can be overcome by IHC staining of target proteins of the tissue slide. IHC provides a selective, high-contrast imaging of cells and tissue components, but their use is largely limited by a significantly more complex laboratory processing and high cost. We proposed a conditional CycleGAN (cCGAN) network to transform the H&E stained images into IHC stained images, facilitating virtual IHC staining on the same slide. This data-driven method requires only a limited amount of labelled data but will generate pixel level segmentation results. The proposed cCGAN model improves the original network \cite{zhu_unpaired_2017} by adding category conditions and introducing two structural loss functions, which realize a multi-subdomain translation and improve the translation accuracy as well. % need to give reasons here. Experiments demonstrate that the proposed model outperforms the original method in unpaired image translation with multi-subdomains. We also explore the potential of unpaired images to image translation method applied on other histology images related tasks with different staining techniques.
Tasks
Published	2019-01-13
URL	http://arxiv.org/abs/1901.04059v1
PDF	http://arxiv.org/pdf/1901.04059v1.pdf
PWC	https://paperswithcode.com/paper/gan-based-virtual-re-staining-a-promising
Repo	https://github.com/Zhaoyang-XU/Virtual-Staining
Framework	none

Audio Denoising with Deep Network Priors


Title	Audio Denoising with Deep Network Priors
Authors	Michael Michelashvili, Lior Wolf
Abstract	We present a method for audio denoising that combines processing done in both the time domain and the time-frequency domain. Given a noisy audio clip, the method trains a deep neural network to fit this signal. Since the fitting is only partly successful and is able to better capture the underlying clean signal than the noise, the output of the network helps to disentangle the clean audio from the rest of the signal. The method is completely unsupervised and only trains on the specific audio clip that is being denoised. Our experiments demonstrate favorable performance in comparison to the literature methods, and our code and audio samples are available at https: //github.com/mosheman5/DNP. Index Terms: Audio denoising; Unsupervised learning
Tasks	Audio Denoising
Published	2019-04-16
URL	https://arxiv.org/abs/1904.07612v2
PDF	https://arxiv.org/pdf/1904.07612v2.pdf
PWC	https://paperswithcode.com/paper/audio-denoising-with-deep-network-priors
Repo	https://github.com/mosheman5/DNP
Framework	pytorch

Systematic quantitative analyses reveal the folk-zoological knowledge embedded in folktales


Title	Systematic quantitative analyses reveal the folk-zoological knowledge embedded in folktales
Authors	Yo Nakawake, Kosuke Sato
Abstract	Cultural learning is a unique human capacity essential for a wide range of adaptations. Researchers have argued that folktales have the pedagogical function of transmitting the essential information for the environment. The most important knowledge for foraging and pastoral society is folk-zoological knowledge, such as the predator-prey relationship among wild animals, or between wild and domesticated animals. Here, we analysed the descriptions of the 382 animal folktales using the natural language processing method and descriptive statistics listed in a worldwide tale-type index (Aarne-Thompson-Uther type index). Our analyses suggested that first, the predator-prey relationship frequently appeared in a co-occurrent animal pair within a folktale (e.g., cat and mouse or wolf and pig), and second, the motif of ‘deception’, describing the antagonistic behaviour among animals, appeared relatively higher in ‘wild and domestic animals’ and ‘wild animals’ than other types. Furthermore, the motif of ‘deception’ appeared more frequently in pairs, corresponding to the predator-prey relationship. These results corresponded with the hypothesis that the combination of animal characters and what happens in stories represented relationships in the real world. The present study demonstrated that the combination of quantitative methods and qualitative data broaden our understanding of the evolutionary aspects of human cultures.
Tasks
Published	2019-07-09
URL	https://arxiv.org/abs/1907.03969v2
PDF	https://arxiv.org/pdf/1907.03969v2.pdf
PWC	https://paperswithcode.com/paper/systematic-quantitative-analyses-reveal-the
Repo	https://github.com/satocos135/animalfolktale-analysis
Framework	none

ORL: Reinforcement Learning Benchmarks for Online Stochastic Optimization Problems


Title	ORL: Reinforcement Learning Benchmarks for Online Stochastic Optimization Problems
Authors	Bharathan Balaji, Jordan Bell-Masterson, Enes Bilgin, Andreas Damianou, Pablo Moreno Garcia, Arpit Jain, Runfei Luo, Alvaro Maggiar, Balakrishnan Narayanaswamy, Chun Ye
Abstract	Reinforcement Learning (RL) has achieved state-of-the-art results in domains such as robotics and games. We build on this previous work by applying RL algorithms to a selection of canonical online stochastic optimization problems with a range of practical applications: Bin Packing, Newsvendor, and Vehicle Routing. While there is a nascent literature that applies RL to these problems, there are no commonly accepted benchmarks which can be used to compare proposed approaches rigorously in terms of performance, scale, or generalizability. This paper aims to fill that gap. For each problem we apply both standard approaches as well as newer RL algorithms and analyze results. In each case, the performance of the trained RL policy is competitive with or superior to the corresponding baselines, while not requiring much in the way of domain knowledge. This highlights the potential of RL in real-world dynamic resource allocation problems.
Tasks	Stochastic Optimization
Published	2019-11-24
URL	https://arxiv.org/abs/1911.10641v2
PDF	https://arxiv.org/pdf/1911.10641v2.pdf
PWC	https://paperswithcode.com/paper/orl-reinforcement-learning-benchmarks-for
Repo	https://github.com/awslabs/or-rl-benchmarks
Framework	none

Emergence of Object Segmentation in Perturbed Generative Models


Title	Emergence of Object Segmentation in Perturbed Generative Models
Authors	Adam Bielski, Paolo Favaro
Abstract	We introduce a novel framework to build a model that can learn how to segment objects from a collection of images without any human annotation. Our method builds on the observation that the location of object segments can be perturbed locally relative to a given background without affecting the realism of a scene. Our approach is to first train a generative model of a layered scene. The layered representation consists of a background image, a foreground image and the mask of the foreground. A composite image is then obtained by overlaying the masked foreground image onto the background. The generative model is trained in an adversarial fashion against a discriminator, which forces the generative model to produce realistic composite images. To force the generator to learn a representation where the foreground layer corresponds to an object, we perturb the output of the generative model by introducing a random shift of both the foreground image and mask relative to the background. Because the generator is unaware of the shift before computing its output, it must produce layered representations that are realistic for any such random perturbation. Finally, we learn to segment an image by defining an autoencoder consisting of an encoder, which we train, and the pre-trained generator as the decoder, which we freeze. The encoder maps an image to a feature vector, which is fed as input to the generator to give a composite image matching the original input image. Because the generator outputs an explicit layered representation of the scene, the encoder learns to detect and segment objects. We demonstrate this framework on real images of several object categories.
Tasks	Semantic Segmentation
Published	2019-05-29
URL	https://arxiv.org/abs/1905.12663v2
PDF	https://arxiv.org/pdf/1905.12663v2.pdf
PWC	https://paperswithcode.com/paper/emergence-of-object-segmentation-in-perturbed
Repo	https://github.com/adambielski/perturbed-seg
Framework	pytorch

TensorNetwork on TensorFlow: A Spin Chain Application Using Tree Tensor Networks


Title	TensorNetwork on TensorFlow: A Spin Chain Application Using Tree Tensor Networks
Authors	Ashley Milsted, Martin Ganahl, Stefan Leichenauer, Jack Hidary, Guifre Vidal
Abstract	TensorNetwork is an open source library for implementing tensor network algorithms in TensorFlow. We describe a tree tensor network (TTN) algorithm for approximating the ground state of either a periodic quantum spin chain (1D) or a lattice model on a thin torus (2D), and implement the algorithm using TensorNetwork. We use a standard energy minimization procedure over a TTN ansatz with bond dimension $\chi$, with a computational cost that scales as $O(\chi^4)$. Using bond dimension $\chi \in [32,256]$ we compare the use of CPUs with GPUs and observe significant computational speed-ups, up to a factor of $100$, using a GPU and the TensorNetwork library.
Tasks	Tensor Networks
Published	2019-05-03
URL	https://arxiv.org/abs/1905.01331v1
PDF	https://arxiv.org/pdf/1905.01331v1.pdf
PWC	https://paperswithcode.com/paper/tensornetwork-on-tensorflow-a-spin-chain
Repo	https://github.com/amilsted/unitreet
Framework	none

MixMatch: A Holistic Approach to Semi-Supervised Learning


Title	MixMatch: A Holistic Approach to Semi-Supervised Learning
Authors	David Berthelot, Nicholas Carlini, Ian Goodfellow, Nicolas Papernot, Avital Oliver, Colin Raffel
Abstract	Semi-supervised learning has proven to be a powerful paradigm for leveraging unlabeled data to mitigate the reliance on large labeled datasets. In this work, we unify the current dominant approaches for semi-supervised learning to produce a new algorithm, MixMatch, that works by guessing low-entropy labels for data-augmented unlabeled examples and mixing labeled and unlabeled data using MixUp. We show that MixMatch obtains state-of-the-art results by a large margin across many datasets and labeled data amounts. For example, on CIFAR-10 with 250 labels, we reduce error rate by a factor of 4 (from 38% to 11%) and by a factor of 2 on STL-10. We also demonstrate how MixMatch can help achieve a dramatically better accuracy-privacy trade-off for differential privacy. Finally, we perform an ablation study to tease apart which components of MixMatch are most important for its success.
Tasks	Image Classification, Semi-Supervised Image Classification
Published	2019-05-06
URL	https://arxiv.org/abs/1905.02249v2
PDF	https://arxiv.org/pdf/1905.02249v2.pdf
PWC	https://paperswithcode.com/paper/mixmatch-a-holistic-approach-to-semi
Repo	https://github.com/gan3sh500/mixmatch-pytorch
Framework	pytorch

Degrees of Laziness in Grounding: Effects of Lazy-Grounding Strategies on ASP Solving


Title	Degrees of Laziness in Grounding: Effects of Lazy-Grounding Strategies on ASP Solving
Authors	Richard Taupe, Antonius Weinzierl, Gerhard Friedrich
Abstract	The traditional ground-and-solve approach to Answer Set Programming (ASP) suffers from the grounding bottleneck, which makes large-scale problem instances unsolvable. Lazy grounding is an alternative approach that interleaves grounding with solving and thus uses space more efficiently. The limited view on the search space in lazy grounding poses unique challenges, however, and can have adverse effects on solving performance. In this paper we present a novel characterization of degrees of laziness in grounding for ASP, i.e. of compromises between lazily grounding as little as possible and the traditional full grounding upfront. We investigate how these degrees of laziness compare to each other formally as well as, by means of an experimental analysis using a number of benchmarks, in terms of their effects on solving performance. Our contributions are the introduction of a range of novel lazy grounding strategies, a formal account on their relationships and their correctness, and an investigation of their effects on solving performance. Experiments show that our approach performs significantly better than state-of-the-art lazy grounding in many cases.
Tasks
Published	2019-03-29
URL	https://arxiv.org/abs/1903.12510v2
PDF	https://arxiv.org/pdf/1903.12510v2.pdf
PWC	https://paperswithcode.com/paper/degrees-of-laziness-in-grounding-effects-of
Repo	https://github.com/alpha-asp/Alpha
Framework	none

Automatic Data Augmentation by Learning the Deterministic Policy


Title	Automatic Data Augmentation by Learning the Deterministic Policy
Authors	Yinghuan Shi, Tiexin Qin, Yong Liu, Jiwen Lu, Yang Gao, Dinggang Shen
Abstract	Aiming to produce sufficient and diverse training samples, data augmentation has been demonstrated for its effectiveness in training deep models. Regarding that the criterion of the best augmentation is challenging to define, we in this paper present a novel learning-based augmentation method termed as DeepAugNet, which formulates the final augmented data as a collection of several sequentially augmented subsets. Specifically, the current augmented subset is required to maximize the performance improvement compared with the last augmented subset by learning the deterministic augmentation policy using deep reinforcement learning. By introducing an unified optimization goal, DeepAugNet intends to combine the data augmentation and the deep model training in an end-to-end training manner which is realized by simultaneously training a hybrid architecture of dueling deep Q-learning algorithm and a surrogate deep model. We extensively evaluated our proposed DeepAugNet on various benchmark datasets including Fashion MNIST, CUB, CIFAR-100 and WebCaricature. Compared with the current state-of-the-arts, our method can achieve a significant improvement in small-scale datasets, and a comparable performance in large-scale datasets. Code will be available soon.
Tasks	Data Augmentation, Q-Learning
Published	2019-10-18
URL	https://arxiv.org/abs/1910.08343v2
PDF	https://arxiv.org/pdf/1910.08343v2.pdf
PWC	https://paperswithcode.com/paper/automatic-data-augmentation-by-learning-the
Repo	https://github.com/WonderSeven/DeepAugNet
Framework	none

Exploiting Partial Knowledge in Declarative Domain-Specific Heuristics for ASP


Title	Exploiting Partial Knowledge in Declarative Domain-Specific Heuristics for ASP
Authors	Richard Taupe, Konstantin Schekotihin, Peter Schüller, Antonius Weinzierl, Gerhard Friedrich
Abstract	Domain-specific heuristics are an important technique for solving combinatorial problems efficiently. We propose a novel semantics for declarative specifications of domain-specific heuristics in Answer Set Programming (ASP). Decision procedures that are based on a partial solution are a frequent ingredient of existing domain-specific heuristics, e.g., for placing an item that has not been placed yet in bin packing. Therefore, in our novel semantics negation as failure and aggregates in heuristic conditions are evaluated on a partial solver state. State-of-the-art solvers do not allow such a declarative specification. Our implementation in the lazy-grounding ASP system Alpha supports heuristic directives under this semantics. By that, we also provide the first implementation for incorporating declaratively specified domain-specific heuristics in a lazy-grounding setting. Experiments confirm that the combination of ASP solving with lazy grounding and our novel heuristics can be a vital ingredient for solving industrial-size problems.
Tasks
Published	2019-09-18
URL	https://arxiv.org/abs/1909.08231v1
PDF	https://arxiv.org/pdf/1909.08231v1.pdf
PWC	https://paperswithcode.com/paper/exploiting-partial-knowledge-in-declarative
Repo	https://github.com/alpha-asp/Alpha
Framework	none

RWR-GAE: Random Walk Regularization for Graph Auto Encoders


Title	RWR-GAE: Random Walk Regularization for Graph Auto Encoders
Authors	Vaibhav, Po-Yao Huang, Robert Frederking
Abstract	Node embeddings have become an ubiquitous technique for representing graph data in a low dimensional space. Graph autoencoders, as one of the widely adapted deep models, have been proposed to learn graph embeddings in an unsupervised way by minimizing the reconstruction error for the graph data. However, its reconstruction loss ignores the distribution of the latent representation, and thus leading to inferior embeddings. To mitigate this problem, we propose a random walk based method to regularize the representations learnt by the encoder. We show that the proposed novel enhancement beats the existing state-of-the-art models by a large margin (upto 7.5%) for node clustering task, and achieves state-of-the-art accuracy on the link prediction task for three standard datasets, cora, citeseer and pubmed. Code available at https://github.com/MysteryVaibhav/DW-GAE.
Tasks	Graph Clustering, Link Prediction
Published	2019-08-12
URL	https://arxiv.org/abs/1908.04003v1
PDF	https://arxiv.org/pdf/1908.04003v1.pdf
PWC	https://paperswithcode.com/paper/rwr-gae-random-walk-regularization-for-graph
Repo	https://github.com/MysteryVaibhav/DW-GAE
Framework	pytorch

Defending against adversarial attacks by randomized diversification


Title	Defending against adversarial attacks by randomized diversification
Authors	Olga Taran, Shideh Rezaeifar, Taras Holotyak, Slava Voloshynovskiy
Abstract	The vulnerability of machine learning systems to adversarial attacks questions their usage in many applications. In this paper, we propose a randomized diversification as a defense strategy. We introduce a multi-channel architecture in a gray-box scenario, which assumes that the architecture of the classifier and the training data set are known to the attacker. The attacker does not only have access to a secret key and to the internal states of the system at the test time. The defender processes an input in multiple channels. Each channel introduces its own randomization in a special transform domain based on a secret key shared between the training and testing stages. Such a transform based randomization with a shared key preserves the gradients in key-defined sub-spaces for the defender but it prevents gradient back propagation and the creation of various bypass systems for the attacker. An additional benefit of multi-channel randomization is the aggregation that fuses soft-outputs from all channels, thus increasing the reliability of the final score. The sharing of a secret key creates an information advantage to the defender. Experimental evaluation demonstrates an increased robustness of the proposed method to a number of known state-of-the-art attacks.
Tasks
Published	2019-04-01
URL	http://arxiv.org/abs/1904.00689v1
PDF	http://arxiv.org/pdf/1904.00689v1.pdf
PWC	https://paperswithcode.com/paper/defending-against-adversarial-attacks-by-1
Repo	https://github.com/taranO/defending-adversarial-attacks-by-RD
Framework	tf

Hidden in Plain Sight For Too Long: Using Text Mining Techniques to Shine a Light on Workplace Sexism and Sexual Harassment


Title	Hidden in Plain Sight For Too Long: Using Text Mining Techniques to Shine a Light on Workplace Sexism and Sexual Harassment
Authors	Amir Karami, Suzanne C. Swan, Cynthia Nicole White, Kayla Ford
Abstract	Objective: The goal of this study is to understand how people experience sexism and sexual harassment in the workplace by discovering themes in 2,362 experiences posted on the Everyday Sexism Project’s website everydaysexism.com. Method: This study used both quantitative and qualitative methods. The quantitative method was a computational framework to collect and analyze a large number of workplace sexual harassment experiences. The qualitative method was the analysis of the topics generated by a text mining method. Results: Twenty-three topics were coded and then grouped into three overarching themes from the sex discrimination and sexual harassment literature. The Sex Discrimination theme included experiences in which women were treated unfavorably due to their sex, such as being passed over for promotion, denied opportunities, paid less than men, and ignored or talked over in meetings. The Sex Discrimination and Gender harassment theme included stories about sex discrimination and gender harassment, such as sexist hostility behaviors ranging from insults and jokes invoking misogynistic stereotypes to bullying behavior. The last theme, Unwanted Sexual Attention, contained stories describing sexual comments and behaviors used to degrade women. Unwanted touching was the highest weighted topic, indicating how common it was for website users to endure being touched, hugged or kissed, groped, and grabbed. Conclusions: This study illustrates how researchers can use automatic processes to go beyond the limits of traditional research methods and investigate naturally occurring large scale datasets on the internet to achieve a better understanding of everyday workplace sexism experiences.
Tasks
Published	2019-07-01
URL	https://arxiv.org/abs/1907.00510v1
PDF	https://arxiv.org/pdf/1907.00510v1.pdf
PWC	https://paperswithcode.com/paper/hidden-in-plain-sight-for-too-long-using-text
Repo	https://github.com/amir-karami/WorkSpace_Sexual_Harresment
Framework	none