Shallow2Deep: Restraining Neural Networks Opacity Through Neural Architecture Search

To read the full-text of this research, you can request a copy directly from the authors.


Recently, the Deep Learning (DL) research community has focused on developing efficient and highly performing Neural Networks (NN). Meanwhile, the eXplainable AI (XAI) research community has focused on making Machine Learning (ML) and Deep Learning methods interpretable and transparent, seeking explainability. This work is a preliminary study on the applicability of Neural Architecture Search (NAS) (a sub-field of DL looking for automatic design of NN structures) in XAI. We propose Shallow2Deep, an evolutionary NAS algorithm that exploits local variability to restrain opacity of DL-systems through NN architectures simplification. Shallow2Deep effectively reduces NN complexity – therefore their opacity – while reaching state-of-the-art performances. Unlike its competitors, Shallow2Deep promotes variability of localised structures in NN, helping to reduce NN opacity. The proposed work analyses the role of local variability in NN architectures design, presenting experimental results that show how this feature is actually desirable.

No full-text available

Request Full-text Paper PDF

To read the full-text of this research,
you can request a copy directly from the authors.

ResearchGate has not been able to resolve any citations for this publication.
Full-text available
Recently, artificial intelligence and machine learning in general have demonstrated remarkable performances in many tasks, from image processing to natural language processing, especially with the advent of deep learning (DL). Along with research progress, they have encroached upon many different fields and disciplines. Some of them require high level of accountability and thus transparency, for example, the medical sector. Explanations for machine decisions and predictions are thus needed to justify their reliability. This requires greater interpretability, which often means we need to understand the mechanism underlying the algorithms. Unfortunately, the blackbox nature of the DL is still unresolved, and many machine decisions are still poorly understood. We provide a review on interpretabilities suggested by different research works and categorize them. The different categories show different dimensions in interpretability research, from approaches that provide "obviously" interpretable information to the studies of complex patterns. By applying the same categorization to interpretability in medical research, it is hoped that: 1) clinicians and practitioners can subsequently approach these methods with caution; 2) insight into interpretability will be born with more considerations for medical practices; and 3) initiatives to push forward data-based, mathematically grounded, and technically grounded medical education are encouraged.
Full-text available
Recently, the eXplainable AI (XAI) research community has focused on developing methods making Machine Learning (ML) predictors more interpretable and explainable. Unfortunately, researchers are struggling to converge towards an unambiguous definition of notions such as interpretation, or, explanation—which are often (and mistakenly) used interchangeably. Furthermore, despite the sound metaphors that Multi-Agent System (MAS) could easily provide to address such a challenge, and agent-oriented perspective on the topic is still missing. Thus, this paper proposes an abstract and formal framework for XAI-based MAS, reconciling notions, and results from the literature.
Full-text available
In the last few years, Artificial Intelligence (AI) has achieved a notable momentum that, if harnessed appropriately, may deliver the best of expectations over many application sectors across the field. For this to occur shortly in Machine Learning, the entire community stands in front of the barrier of explainability, an inherent problem of the latest techniques brought by sub-symbolism (e.g. ensembles or Deep Neural Networks) that were not present in the last hype of AI (namely, expert systems and rule based models). Paradigms underlying this problem fall within the so-called eXplainable AI (XAI) field, which is widely acknowledged as a crucial feature for the practical deployment of AI models. The overview presented in this article examines the existing literature and contributions already done in the field of XAI, including a prospect toward what is yet to be reached. For this purpose we summarize previous efforts made to define explainability in Machine Learning, establishing a novel definition of explainable Machine Learning that covers such prior conceptual propositions with a major focus on the audience for which the explainability is sought. Departing from this definition, we propose and discuss about a taxonomy of recent contributions related to the explainability of different Machine Learning models, including those aimed at explaining Deep Learning methods for which a second dedicated taxonomy is built and examined in detail. This critical literature analysis serves as the motivating background for a series of challenges faced by XAI, such as the interesting crossroads of data fusion and explainability. Our prospects lead toward the concept of Responsible Artificial Intelligence , namely, a methodology for the large-scale implementation of AI methods in real organizations with fairness, model explainability and accountability at its core. Our ultimate goal is to provide newcomers to the field of XAI with a thorough taxonomy that can serve as reference material in order to stimulate future research advances, but also to encourage experts and professionals from other disciplines to embrace the benefits of AI in their activity sectors, without any prior bias for its lack of interpretability.
Full-text available
At the dawn of the fourth industrial revolution, we are witnessing a fast and widespread adoption of Artificial Intelligence (AI) in our daily life, which contributes to accelerating the shift towards a more algorithmic society. However, even with such unprecedented advancements, a key impediment to the use of AI-based systems is that they often lack transparency. Indeed, the black box nature of these systems allows powerful predictions, but it cannot be directly explained. This issue has triggered a new debate on Explainable Artificial Intelligence. A research field that holds substantial promise for improving trust and transparency of AI-based systems. It is recognized as the sine qua non for AI to continue making steady progress without disruption. This survey provides an entry point for interested researchers and practitioners to learn key aspects of the young and rapidly growing body of research related to explainable AI. Through the lens of literature, we review existing approaches regarding the topic, we discuss trends surrounding its sphere and we present major research trajectories.
Full-text available
Supervised machine-learning models boast remarkable predictive capabilities. But can you trust your model? Will it work in deployment? What else can it tell you about the world? Models should be not only good, but also interpretable, yet the task of interpretation appears underspecified. The academic literature has provided diverse and sometimes non-overlapping motivations for interpretability and has offered myriad techniques for rendering interpretable models. Despite this ambiguity, many authors proclaim their models to be interpretable axiomatically, absent further argument. Problematically, it is not clear what common properties unite these techniques. This article seeks to refine the discourse on interpretability. First it examines the objectives of previous papers addressing interpretability, finding them to be diverse and occasionally discordant. Then, it explores model properties and techniques thought to confer interpretability, identifying transparency to humans and post hoc explanations as competing concepts. Throughout, the feasibility and desirability of different notions of interpretability are discussed. The article questions the oft-made assertions that linear models are interpretable and that deep neural networks are not. © 2018 Association for Computing Machinery. All rights reserved.
Full-text available
Conference Paper
In the last decade, with availability of large datasets and more computing power, machine learning systems have achieved (super)human performance in a wide variety of tasks. Examples of this rapid development can be seen in image recognition, speech analysis, strategic game planning and many more. The problem with many state-of-the-art models is a lack of transparency and interpretability. The lack of thereof is a major drawback in many applications, e.g. healthcare and finance, where rationale for model's decision is a requirement for trust. In the light of these issues, explainable artificial intelligence (XAI) has become an area of interest in research community. This paper summarizes recent developments in XAI in supervised learning, starts a discussion on its connection with artificial general intelligence, and gives proposals for further research directions.
Full-text available
In the last years many accurate decision support systems have been constructed as black boxes, that is as systems that hide their internal logic to the user. This lack of explanation constitutes both a practical and an ethical issue. The literature reports many approaches aimed at overcoming this crucial weakness sometimes at the cost of scarifying accuracy for interpretability. The applications in which black box decision systems can be used are various, and each approach is typically developed to provide a solution for a specific problem and, as a consequence, delineating explicitly or implicitly its own definition of interpretability and explanation. The aim of this paper is to provide a classification of the main problems addressed in the literature with respect to the notion of explanation and the type of black box system. Given a problem definition, a black box type, and a desired explanation this survey should help the researcher to find the proposals more useful for his own work. The proposed classification of approaches to open black box models should also be useful for putting the many research open questions in perspective.
Full-text available
We present Fashion-MNIST, a new dataset comprising of 28x28 grayscale images of 70,000 fashion products from 10 categories, with 7,000 images per category. The training set has 60,000 images and the test set has 10,000 images. Fashion-MNIST is intended to serve as a direct drop-in replacement for the original MNIST dataset for benchmarking machine learning algorithms, as it shares the same image size, data format and the structure of training and testing splits. The dataset is freely available at
Full-text available
We present a class of efficient models called MobileNets for mobile and embedded vision applications. MobileNets are based on a streamlined architecture that uses depth-wise separable convolutions to build light weight deep neural networks. We introduce two simple global hyper-parameters that efficiently trade off between latency and accuracy. These hyper-parameters allow the model builder to choose the right sized model for their application based on the constraints of the problem. We present extensive experiments on resource and accuracy tradeoffs and show strong performance compared to other popular models on ImageNet classification. We then demonstrate the effectiveness of MobileNets across a wide range of applications and use cases including object detection, finegrain classification, face attributes and large scale geo-localization.
Full-text available
Conference Paper
At present the deep neural network is the hottest topic in the domain of machine learning and can accomplish a deep hierarchical representation of the input data. Due to deep architecture the large convolutional neural networks can reach very small test error rates below 0.4% using the MNIST database. In this work we have shown, that high accuracy can be achieved using reduced shallow convolutional neural network without adding distortions for digits. The main contribution of this paper is to point out how using simplified convolutional neural network is to obtain test error rate 0.71% on the MNIST handwritten digit benchmark. It permits to reduce computational resources in order to model convolutional neural network.
Full-text available
Deep neural networks (DNNs) have demonstrated impressive performance in complex machine learning tasks such as image classification or speech recognition. However, due to their multilayer nonlinear structure, they are not transparent, i.e., it is hard to grasp what makes them arrive at a particular classification or recognition decision, given a new unseen data sample. Recently, several approaches have been proposed enabling one to understand and interpret the reasoning embodied in a DNN for a single test image. These methods quantify the "importance" of individual pixels with respect to the classification decision and allow a visualization in terms of a heatmap in pixel/input space. While the usefulness of heatmaps can be judged subjectively by a human, an objective quality measure is missing. In this paper, we present a general methodology based on region perturbation for evaluating ordered collections of pixels such as heatmaps. We compare heatmaps computed by three different methods on the SUN397, ILSVRC2012, and MIT Places data sets. Our main result is that the recently proposed layer-wise relevance propagation algorithm qualitatively and quantitatively provides a better explanation of what made a DNN arrive at a particular classification decision than the sensitivity-based approach or the deconvolution method. We provide theoretical arguments to explain this result and discuss its practical implications. Finally, we investigate the use of heatmaps for unsupervised assessment of the neural network performance.
Full-text available
We can better understand deep neural networks by identifying which features each of their neurons have learned to detect. To do so, researchers have created Deep Visualization techniques including activation maximization, which synthetically generates inputs (e.g. images) that maximally activate each neuron. A limitation of current techniques is that they assume each neuron detects only one type of feature, but we know that neurons can be multifaceted, in that they fire in response to many different types of features: for example, a grocery store class neuron must activate either for rows of produce or for a storefront. Previous activation maximization techniques constructed images without regard for the multiple different facets of a neuron, creating inappropriate mixes of colors, parts of objects, scales, orientations, etc. Here, we introduce an algorithm that explicitly uncovers the multiple facets of each neuron by producing a synthetic visualization of each of the types of images that activate a neuron. We also introduce regularization methods that produce state-of-the-art results in terms of the interpretability of images obtained by activation maximization. By separately synthesizing each type of image a neuron fires in response to, the visualizations have more appropriate colors and coherent global structure. Multifaceted feature visualization thus provides a clearer and more comprehensive description of the role of each neuron.
Full-text available
The field of machine learning has taken a dramatic twist in recent times, with the rise of the Artificial Neural Network (ANN). These biologically inspired computational models are able to far exceed the performance of previous forms of artificial intelligence in common machine learning tasks. One of the most impressive forms of ANN architecture is that of the Convolutional Neural Network (CNN). CNNs are primarily used to solve difficult image-driven pattern recognition tasks and with their precise yet simple architecture, offers a simplified method of getting started with ANNs. This document provides a brief introduction to CNNs, discussing recently published papers and newly formed techniques in developing these brilliantly fantastic image recognition models. This introduction assumes you are familiar with the fundamentals of ANNs and machine learning.
Full-text available
Recent years have produced great advances in training large, deep neural networks (DNNs), including notable successes in training convolutional neural networks (convnets) to recognize natural images. However, our understanding of how these models work, especially what computations they perform at intermediate layers, has lagged behind. Progress in the field will be further accelerated by the development of better tools for visualizing and interpreting neural nets. We introduce two such tools here. The first is a tool that visualizes the activations produced on each layer of a trained convnet as it processes an image or video (e.g. a live webcam stream). We have found that looking at live activations that change in response to user input helps build valuable intuitions about how convnets work. The second tool enables visualizing features at each layer of a DNN via regularized optimization in image space. Because previous versions of this idea produced less recognizable images, here we introduce several new regularization methods that combine to produce qualitatively clearer, more interpretable visualizations. Both tools are open source and work on a pre-trained convnet with minimal setup.
Full-text available
Understanding and interpreting classification decisions of automated image classification systems is of high value in many applications as it allows to verify the reasoning of the system and provides additional information to the human expert. Although machine learning methods are solving very successfully a plethora of tasks, they have in most cases the disadvantage of acting as a black box, not providing any information about what made them arrive at a particular decision. This work proposes a general solution to the problem of understanding classification decisions by pixel-wise decomposition of non- linear classifiers. We introduce a methodology that allows to visualize the contributions of single pixels to predictions for kernel-based classifiers over Bag of Words features and for multilayered neural networks. These pixel contributions can be visualized as heatmaps and are provided to a human expert who can intuitively not only verify the validity of the classification decision, but also focus further analysis on regions of potential interest. We evaluate our method for classifiers trained on PASCAL VOC 2009 images, synthetic image data containing geometric shapes, the MNIST handwritten digits data set and for the pre-trained ImageNet model available as part of the Caffe open source package.
Full-text available
Conference Paper
The explosion of image data on the Internet has the potential to foster more sophisticated and robust models and algorithms to index, retrieve, organize and interact with images and multimedia data. But exactly how such data can be harnessed and organized remains a critical problem. We introduce here a new database called ldquoImageNetrdquo, a large-scale ontology of images built upon the backbone of the WordNet structure. ImageNet aims to populate the majority of the 80,000 synsets of WordNet with an average of 500-1000 clean and full resolution images. This will result in tens of millions of annotated images organized by the semantic hierarchy of WordNet. This paper offers a detailed analysis of ImageNet in its current state: 12 subtrees with 5247 synsets and 3.2 million images in total. We show that ImageNet is much larger in scale and diversity and much more accurate than the current image datasets. Constructing such a large-scale database is a challenging task. We describe the data collection scheme with Amazon Mechanical Turk. Lastly, we illustrate the usefulness of ImageNet through three simple applications in object recognition, image classification and automatic object clustering. We hope that the scale, accuracy, diversity and hierarchical structure of ImageNet can offer unparalleled opportunities to researchers in the computer vision community and beyond.
Fabricating neural models for a wide range of mobile devices is a challenging task due to highly constrained resources. Recent trends favor neural architecture search involving evolutionary algorithms (EA) and reinforcement learning (RL), however, they are separately used. In this paper, we present a novel multi-objective algorithm called MoreMNAS (Multi-Objective Reinforced Evolution in Mobile Neural Architecture Search) by leveraging good virtues from both sides. Particularly, we devise a variant of multi-objective genetic algorithm NSGA-II, where mutations are performed either by reinforcement learning or a natural mutating process. It maintains a delicate balance between exploration and exploitation. Not only does it prevent the search degradation, but it also makes use of the learned knowledge. Our experiments conducted in Super-resolution domain (SR) deliver rivaling models compared with some state-of-the-art methods at fewer FLOPS (Evaluation code can be found at
The more intelligent systems based on sub-symbolic techniques pervade our everyday lives, the less human can understand them. This is why symbolic approaches are getting more and more attention in the general effort to make AI interpretable, explainable, and trustable. Understanding the current state of the art of AI techniques integrating symbolic and sub-symbolic approaches is then of paramount importance, nowadays-in particular in the XAI perspective. This is why this paper provides an overview of the main symbolic/sub-symbolic integration techniques, focussing in particular on those targeting explainable AI systems.
How to build a high-quality multi-domain dialogue system is a challenging work due to its complicated and entangled dialogue state space among each domain, which seriously limits the quality of dialogue policy, and further affects the generated response. In this paper, we propose a novel method to acquire a satisfying policy and subtly circumvent the knotty dialogue state representation problem in the multi-domain setting. Inspired by real school teaching scenarios, our method is composed of multiple domain-specific teachers and a universal student. Each individual teacher only focuses on one specific domain and learns its corresponding domain knowledge and dialogue policy based on a precisely extracted single domain dialogue state representation. Then, these domain-specific teachers impart their domain knowledge and policies to a universal student model and collectively make this student model a multi-domain dialogue expert. Experiment results show that our method reaches competitive results with SOTAs in both multi-domain and single domain setting.
The effort devoted to hand-crafting image classifiers has motivated the use of architecture search to discover them automatically. Reinforcement learning and evolution have both shown promise for this purpose. This study introduces a regularized version of a popular asynchronous evolutionary algorithm. We rigorously compare it to the non-regularized form and to a highly-successful reinforcement learning baseline. Using the same hardware, compute effort and neural network training code, we conduct repeated experiments side-by-side, exploring different datasets, search spaces and scales. We show regularized evolution consistently produces models with similar or higher accuracy, across a variety of contexts without need for re-tuning parameters. In addition, regularized evolution exhibits considerably better performance than reinforcement learning at early search stages, suggesting it may be the better choice when fewer compute resources are available. This constitutes the first controlled comparison of the two search algorithms in this context. Finally, we present new architectures discovered with regularized evolution that we nickname AmoebaNets. These models set a new state of the art for CIFAR-10 (mean test error = 2.13%) and mobile-size ImageNet (top-5 accuracy = 92.1% with 5.06M parameters), and reach the current state of the art for ImageNet (top-5 accuracy = 96.2%).
We propose a method for learning CNN structures that is more efficient than previous approaches: instead of using reinforcement learning (RL) or genetic algorithms (GA), we use a sequential model-based optimization (SMBO) strategy, in which we search for architectures in order of increasing complexity, while simultaneously learning a surrogate function to guide the search, similar to A* search. On the CIFAR-10 dataset, our method finds a CNN structure with the same classification accuracy (3.41% error rate) as the RL method of Zoph et al. (2017), but 2 times faster (in terms of number of models evaluated). It also outperforms the GA method of Liu et al. (2017), which finds a model with worse performance (3.63% error rate), and takes 5 times longer. Finally we show that the model we learned on CIFAR also works well at the task of ImageNet classification. In particular, we match the state-of-the-art performance of 82.9% top-1 and 96.1% top-5 accuracy.
It is well-known that neural networks are universal approximators, but that deeper networks tend to be much more efficient than shallow ones. We shed light on this by proving that the total number of neurons $m$ required to approximate natural classes of multivariate polynomials of $n$ variables grows only linearly with $n$ for deep neural networks, but grows exponentially when merely a single hidden layer is allowed. We also provide evidence that when the number of hidden layers is increased from $1$ to $k$, the neuron requirement grows exponentially not with $n$ but with $n^{1/k}$, suggesting that the minimum number of layers required for computational tractability grows only logarithmically with $n$.
While convolutional neural network (CNN) architectures have achieved great success in various vision tasks, the critical scale problem is still much under-explored, especially for pedestrian detection. Current approaches mainly focus on using large numbers of training images with different scales to improve the network capability or result fusions by multi-scale crops of images during testing. Designing a CNN architecture that can intrinsically capture the characteristics of large-scale and small-scale objects and also retain the scale invariance property is still a very challenging problem. In this paper, we propose a novel scale-aware Fast R-CNN to handle the detection of small object instances which are very common in pedestrian detection. Our architecture incorporates a large-scale sub-network and a small-scale sub-network into a unified architecture by leveraging the scale-aware weighting during training. The heights of object proposals are utilized to specify different scale-aware weights for the two sub-networks. Extensive evaluations on the challenging Caltech~\cite{dollar2012pedestrian} demonstrate the superiority of the proposed architecture over the state-of-the-art methods~\cite{compact,ta_cnn}. In particular, the miss rate on the Caltech dataset is reduced to $9.68\%$ by our method, significantly smaller than $11.75\%$ by CompACT-Deep~\cite{compact} and $20.86\%$ by TA-CNN~\cite{ta_cnn}.
Most object detectors contain two important components: a feature extractor and an object classifier. The feature extractor has rapidly evolved with significant research efforts leading to better deep ConvNet architectures. The object classifier, however, has not received much attention and most state-of-the-art systems (like R-CNN) use simple multi-layer perceptrons. This paper demonstrates that carefully designing deep networks for object classification is just as important. We take inspiration from traditional object classifiers, such as DPM, and experiment with deep networks that have part-like filters and reason over latent variables. We discover that on pre-trained convolutional feature maps, even randomly initialized deep classifiers produce excellent results, while the improvement due to fine-tuning is secondary; on HOG features, deep classifiers outperform DPMs and produce the best HOG-only results without external data. We believe these findings provide new insight for developing object detection systems. Our framework, called Networks on Convolutional feature maps (NoC), achieves outstanding results on the PASCAL VOC 2007 (73.3% mAP) and 2012 (68.8% mAP) benchmarks.
Neural networks (NNs) have been successfully applied to solve a variety of application problems including classification and function approximation. They are especially useful as function approximators because they do not require prior knowledge of the input data distribution and they have been shown to be universal approximators. In many applications, it is desirable to extract knowledge that can explain how Me problems are solved by the networks. Most existing approaches have focused on extracting symbolic rules for classification. Few methods have been devised to extract rules from trained NNs for regression. This article presents an approach for extracting rules from trained NNs for regression. Each rule in the extracted rule set corresponds to a subregion of the input space and a linear function involving the relevant input attributes of the data approximates the network output for all data samples in this subregion. Extensive experimental results on 32 benchmark data sets demonstrate the effectiveness of the proposed approach in generating accurate regression rules
Progressive neural architecture search
  • C Liu
  • V Ferrari
  • M Hebert
  • C Sminchisescu
  • Y Weiss
Proxylessnas: Direct neural architecture search on target task and hardware
  • H Cai
  • L Zhu
  • S Han
Cai, H., Zhu, L., Han, S.: Proxylessnas: Direct neural architecture search on target task and hardware. In: 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. (2019),
Probabilistic neural architecture search
  • F P Casale
  • J Gordon
  • N Fusi
Casale, F.P., Gordon, J., Fusi, N.: Probabilistic neural architecture search. CoRR abs/1902.05116 (2019),
Theory of the backpropagation neural network
  • R Hecht-Nielsen
Hecht-Nielsen, R.: Theory of the backpropagation neural network. Neural Networks 1(Supplement-1), 445-448 (1988).
CIFAR10 to compare visual recognition performance between deep neural networks and humans
  • T Ho-Phuoc
Ho-Phuoc, T.: CIFAR10 to compare visual recognition performance between deep neural networks and humans. CoRR abs/1811.07270 (2018), abs/1811.07270
Feature relevance quantification in explainable AI: A causal problem
  • D Janzing
  • L Minorics
  • P Blöbaum
Janzing, D., Minorics, L., Blöbaum, P.: Feature relevance quantification in explainable AI: A causal problem. In: Chiappa, S., Calandra, R. (eds.) The 23rd International Conference on Artificial Intelligence and Statistics, AISTATS 2020, 26-28 August 2020, Online [Palermo, Sicily, Italy]. Proceedings of Machine Learning Research, vol. 108, pp. 2907-2916. PMLR (2020), http://proceedings.mlr. press/v108/janzing20a.html
Shallow-deep networks: Understanding and mitigating network overthinking
  • Y Kaya
  • S Hong
  • T Dumitras
Kaya, Y., Hong, S., Dumitras, T.: Shallow-deep networks: Understanding and mitigating network overthinking. In: Chaudhuri, K., Salakhutdinov, R. (eds.) Proceedings of the 36th International Conference on Machine Learning, ICML 2019, 9-15 June 2019, Long Beach, California, USA. Proceedings of Machine Learning Research, vol. 97, pp. 3301-3310. PMLR (2019), v97/kaya19a.html
The mythos of model interpretability
  • Z C Lipton
Lipton, Z.C.: The mythos of model interpretability. Queue 16(3), 31-57 (Jun 2018).
DARTS: differentiable architecture search
  • H Liu
  • K Simonyan
  • Y Yang
Liu, H., Simonyan, K., Yang, Y.: DARTS: differentiable architecture search. In: 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019. (2019), https://openreview. net/forum?id=S1eYHoC5FX
Neural architecture search with GBDT
  • R Luo
  • X Tan
  • R Wang
  • T Qin
  • E Chen
  • T Liu
Luo, R., Tan, X., Wang, R., Qin, T., Chen, E., Liu, T.: Neural architecture search with GBDT. CoRR abs/2007.04785 (2020), 2007.04785
Synthesizing the preferred inputs for neurons in neural networks via deep generator networks
  • A M Nguyen
  • A Dosovitskiy
  • J Yosinski
  • T Brox
  • J Clune
  • D D Lee
  • M Sugiyama
  • U Von Luxburg
  • I Guyon
Nguyen, A.M., Dosovitskiy, A., Yosinski, J., Brox, T., Clune, J.: Synthesizing the preferred inputs for neurons in neural networks via deep generator networks. In: Lee, D.D., Sugiyama, M., von Luxburg, U., Guyon, I., Garnett, R. (eds.) Advances in Neural Information Processing Systems 29: Annual Conference on Neural Information Processing Systems 2016, December 5-10, 2016, Barcelona, Spain. pp. 3387-3395 (2016), hash/5d79099fcdf499f12b79770834c0164a-Abstract.html
Very deep convolutional networks for large-scale image recognition
  • K Simonyan
  • A Zisserman
Simonyan, K., Zisserman, A.: Very deep convolutional networks for large-scale image recognition. In: Bengio, Y., LeCun, Y. (eds.) 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings (2015),
Efficientnet: Rethinking model scaling for convolutional neural networks
  • M Tan
  • Q V Le
Tan, M., Le, Q.V.: Efficientnet: Rethinking model scaling for convolutional neural networks. CoRR abs/1905.11946 (2019),
A survey on neural architecture search
  • M Wistuba
  • A Rawat
  • T Pedapati
Wistuba, M., Rawat, A., Pedapati, T.: A survey on neural architecture search. CoRR abs/1905.01392 (2019),