Surrogate-Assisted Evolutionary Deep Learning Using an End-to-End Random Forest-Based Performance Predictor

© 1997-2012 IEEE. Convolutional neural networks (CNNs) have shown remarkable performance in various real-world applications. Unfortunately, the promising performance of CNNs can be achieved only when their architectures are optimally constructed. The architectures of state-of-the-art CNNs are typically handcrafted with extensive expertise in both CNNs and the investigated data, which consequently hampers the widespread adoption of CNNs for less experienced users. Evolutionary deep learning (EDL) is able to automatically design the best CNN architectures without much expertise. However, the existing EDL algorithms generally evaluate the fitness of a new architecture by training from scratch, resulting in the prohibitive computational cost even operated on high-performance computers. In this paper, an end-to-end offline performance predictor based on the random forest is proposed to accelerate the fitness evaluation in EDL. The proposed performance predictor shows the promising performance in term of the classification accuracy and the consumed computational resources when compared with 18 state-of-the-art peer competitors by integrating into an existing EDL algorithm as a case study. The proposed performance predictor is also compared with the other two representatives of existing performance predictors. The experimental results show the proposed performance predictor not only significantly speeds up the fitness evaluations but also achieves the best prediction among the peer performance predictors.

Download Full-text

Surrogate-Assisted Evolutionary Deep Learning Using an End-to-End Random Forest-Based Performance Predictor

10.26686/wgtn.13158302 ◽

2020 ◽

Author(s):

Y Sun ◽

H Wang ◽

Bing Xue ◽

Y Jin ◽

GG Yen ◽

...

Keyword(s):

Deep Learning ◽

Random Forest ◽

High Performance ◽

State Of The Art ◽

Computational Cost ◽

Fitness Evaluation ◽

Performance Predictors ◽

End To End ◽

Computational Resources ◽

High Performance Computers

Download Full-text

SHEDR: An End-to-End Deep Neural Event Detection and Recommendation Framework for Hyperlocal News Using Social Media

INFORMS Journal on Computing ◽

10.1287/ijoc.2021.1112 ◽

2021 ◽

Author(s):

Yuheng Hu ◽

Yili Hong

Keyword(s):

Neural Network ◽

Social Media ◽

Deep Learning ◽

Event Detection ◽

Large Scale ◽

Short Term Memory ◽

State Of The Art ◽

Neural Network Models ◽

Neural Event ◽

End To End

Residents often rely on newspapers and television to gather hyperlocal news for community awareness and engagement. More recently, social media have emerged as an increasingly important source of hyperlocal news. Thus far, the literature on using social media to create desirable societal benefits, such as civic awareness and engagement, is still in its infancy. One key challenge in this research stream is to timely and accurately distill information from noisy social media data streams to community members. In this work, we develop SHEDR (social media–based hyperlocal event detection and recommendation), an end-to-end neural event detection and recommendation framework with a particular use case for Twitter to facilitate residents’ information seeking of hyperlocal events. The key model innovation in SHEDR lies in the design of the hyperlocal event detector and the event recommender. First, we harness the power of two popular deep neural network models, the convolutional neural network (CNN) and long short-term memory (LSTM), in a novel joint CNN-LSTM model to characterize spatiotemporal dependencies for capturing unusualness in a region of interest, which is classified as a hyperlocal event. Next, we develop a neural pairwise ranking algorithm for recommending detected hyperlocal events to residents based on their interests. To alleviate the sparsity issue and improve personalization, our algorithm incorporates several types of contextual information covering topic, social, and geographical proximities. We perform comprehensive evaluations based on two large-scale data sets comprising geotagged tweets covering Seattle and Chicago. We demonstrate the effectiveness of our framework in comparison with several state-of-the-art approaches. We show that our hyperlocal event detection and recommendation models consistently and significantly outperform other approaches in terms of precision, recall, and F-1 scores. Summary of Contribution: In this paper, we focus on a novel and important, yet largely underexplored application of computing—how to improve civic engagement in local neighborhoods via local news sharing and consumption based on social media feeds. To address this question, we propose two new computational and data-driven methods: (1) a deep learning–based hyperlocal event detection algorithm that scans spatially and temporally to detect hyperlocal events from geotagged Twitter feeds; and (2) A personalized deep learning–based hyperlocal event recommender system that systematically integrates several contextual cues such as topical, geographical, and social proximity to recommend the detected hyperlocal events to potential users. We conduct a series of experiments to examine our proposed models. The outcomes demonstrate that our algorithms are significantly better than the state-of-the-art models and can provide users with more relevant information about the local neighborhoods that they live in, which in turn may boost their community engagement.

Download Full-text

An Experiment on the Use of Genetic Algorithms for Topology Selection in Deep Learning

Journal of Electrical and Computer Engineering ◽

10.1155/2019/3217542 ◽

2019 ◽

Vol 2019 ◽

pp. 1-12 ◽

Cited By ~ 1

Author(s):

Fernando Mattioli ◽

Daniel Caetano ◽

Alexandre Cardoso ◽

Eduardo Naves ◽

Edgard Lamounier

Keyword(s):

Genetic Algorithms ◽

Deep Learning ◽

Deep Neural Networks ◽

State Of The Art ◽

Complex Task ◽

Trial And Error ◽

Complex Solution ◽

Computation Algorithms ◽

Computational Resources ◽

Topology Selection

The choice of a good topology for a deep neural network is a complex task, essential for any deep learning project. This task normally demands knowledge from previous experience, as the higher amount of required computational resources makes trial and error approaches prohibitive. Evolutionary computation algorithms have shown success in many domains, by guiding the exploration of complex solution spaces in the direction of the best solutions, with minimal human intervention. In this sense, this work presents the use of genetic algorithms in deep neural networks topology selection. The evaluated algorithms were able to find competitive topologies while spending less computational resources when compared to state-of-the-art methods.

Download Full-text

Repetitive Reprediction Deep Decipher for Semi-Supervised Learning

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v34i04.6082 ◽

2020 ◽

Vol 34 (04) ◽

pp. 6170-6177

Author(s):

Guo-Hua Wang ◽

Jianxin Wu

Keyword(s):

Deep Learning ◽

Supervised Learning ◽

Large Scale ◽

State Of The Art ◽

Link Function ◽

Training Strategy ◽

Network Parameters ◽

Theoretical Support ◽

Percentage Points ◽

End To End

Most recent semi-supervised deep learning (deep SSL) methods used a similar paradigm: use network predictions to update pseudo-labels and use pseudo-labels to update network parameters iteratively. However, they lack theoretical support and cannot explain why predictions are good candidates for pseudo-labels. In this paper, we propose a principled end-to-end framework named deep decipher (D2) for SSL. Within the D2 framework, we prove that pseudo-labels are related to network predictions by an exponential link function, which gives a theoretical support for using predictions as pseudo-labels. Furthermore, we demonstrate that updating pseudo-labels by network predictions will make them uncertain. To mitigate this problem, we propose a training strategy called repetitive reprediction (R2). Finally, the proposed R2-D2 method is tested on the large-scale ImageNet dataset and outperforms state-of-the-art methods by 5 percentage points.

Download Full-text

High Performance Gesture Recognition via Effective and Efficient Temporal Modeling

Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2019/141 ◽

2019 ◽

Author(s):

Yang Yi ◽

Feng Ni ◽

Yuexin Ma ◽

Xinge Zhu ◽

Yuankai Qi ◽

...

Keyword(s):

Neural Networks ◽

Convolutional Neural Networks ◽

Gesture Recognition ◽

High Performance ◽

Short Term Memory ◽

State Of The Art ◽

Computational Cost ◽

Temporal Modeling ◽

Spatiotemporal Features ◽

Public Datasets

State-of-the-art hand gesture recognition methods have investigated the spatiotemporal features based on 3D convolutional neural networks (3DCNNs) or convolutional long short-term memory (ConvLSTM). However, they often suffer from the inefficiency due to the high computational complexity of their network structures. In this paper, we focus instead on the 1D convolutional neural networks and propose a simple and efficient architectural unit, Multi-Kernel Temporal Block (MKTB), that models the multi-scale temporal responses by explicitly applying different temporal kernels. Then, we present a Global Refinement Block (GRB), which is an attention module for shaping the global temporal features based on the cross-channel similarity. By incorporating the MKTB and GRB, our architecture can effectively explore the spatiotemporal features within tolerable computational cost. Extensive experiments conducted on public datasets demonstrate that our proposed model achieves the state-of-the-art with higher efficiency. Moreover, the proposed MKTB and GRB are plug-and-play modules and the experiments on other tasks, like video understanding and video-based person re-identification, also display their good performance in efficiency and capability of generalization.

Download Full-text

SEVEN: Deep Semi-supervised Verification Networks

Proceedings of the Twenty-Sixth International Joint Conference on Artificial Intelligence ◽

10.24963/ijcai.2017/358 ◽

2017 ◽

Cited By ~ 4

Author(s):

Vahid Noroozi ◽

Lei Zheng ◽

Sara Bahaadini ◽

Sihong Xie ◽

Philip S. Yu

Keyword(s):

Deep Learning ◽

State Of The Art ◽

Learning Models ◽

Fingerprint Verification ◽

Short Supply ◽

Two Samples ◽

End To End

Verification determines whether two samples belong to the same class or not, and has important applications such as face and fingerprint verification, where thousands or millions of categories are present but each category has scarce labeled examples, presenting two major challenges for existing deep learning models. We propose a deep semi-supervised model named SEmi-supervised VErification Network (SEVEN) to address these challenges. The model consists of two complementary components. The generative component addresses the lack of supervision within each category by learning general salient structures from a large amount of data across categories. The discriminative component exploits the learned general features to mitigate the lack of supervision within categories, and also directs the generative component to find more informative structures of the whole data manifold. The two components are tied together in SEVEN to allow an end-to-end training of the two components. Extensive experiments on four verification tasks demonstrate that SEVEN significantly outperforms other state-of-the-art deep semi-supervised techniques when labeled data are in short supply. Furthermore, SEVEN is competitive with fully supervised baselines trained with a larger amount of labeled data. It indicates the importance of the generative component in SEVEN.

Download Full-text

End-to-end, sequence-to-sequence probabilistic visual odometry through deep neural networks

The International Journal of Robotics Research ◽

10.1177/0278364917734298 ◽

2017 ◽

Vol 37 (4-5) ◽

pp. 513-542 ◽

Cited By ~ 41

Author(s):

Sen Wang ◽

Ronald Clark ◽

Hongkai Wen ◽

Niki Trigoni

Keyword(s):

Neural Networks ◽

Deep Learning ◽

Convolutional Neural Networks ◽

Deep Neural Networks ◽

State Of The Art ◽

Visual Odometry ◽

Feature Representation ◽

Geometric Information ◽

Geometric Problem ◽

End To End

This paper studies visual odometry (VO) from the perspective of deep learning. After tremendous efforts in the robotics and computer vision communities over the past few decades, state-of-the-art VO algorithms have demonstrated incredible performance. However, since the VO problem is typically formulated as a pure geometric problem, one of the key features still missing from current VO systems is the capability to automatically gain knowledge and improve performance through learning. In this paper, we investigate whether deep neural networks can be effective and beneficial to the VO problem. An end-to-end, sequence-to-sequence probabilistic visual odometry (ESP-VO) framework is proposed for the monocular VO based on deep recurrent convolutional neural networks. It is trained and deployed in an end-to-end manner, that is, directly inferring poses and uncertainties from a sequence of raw images (video) without adopting any modules from the conventional VO pipeline. It can not only automatically learn effective feature representation encapsulating geometric information through convolutional neural networks, but also implicitly model sequential dynamics and relation for VO using deep recurrent neural networks. Uncertainty is also derived along with the VO estimation without introducing much extra computation. Extensive experiments on several datasets representing driving, flying and walking scenarios show competitive performance of the proposed ESP-VO to the state-of-the-art methods, demonstrating a promising potential of the deep learning technique for VO and verifying that it can be a viable complement to current VO systems.

Download Full-text

A Layer Decomposition-Recomposition Framework for Neuron Pruning towards Accurate Lightweight Networks

Proceedings of the AAAI Conference on Artificial Intelligence ◽

10.1609/aaai.v33i01.33013355 ◽

2019 ◽

Vol 33 ◽

pp. 3355-3362 ◽

Cited By ~ 1

Author(s):

Weijie Chen ◽

Yuan Zhang ◽

Di Xie ◽

Shiliang Pu

Keyword(s):

State Of The Art ◽

Computational Cost ◽

Fine Tuning ◽

Layer By Layer ◽

Storage Overhead ◽

Speed Up ◽

Output Neurons ◽

End To End ◽

Output Information ◽

And Storage

Neuron pruning is an efficient method to compress the network into a slimmer one for reducing the computational cost and storage overhead. Most of state-of-the-art results are obtained in a layer-by-layer optimization mode. It discards the unimportant input neurons and uses the survived ones to reconstruct the output neurons approaching to the original ones in a layer-by-layer manner. However, an unnoticed problem arises that the information loss is accumulated as layer increases since the survived neurons still do not encode the entire information as before. A better alternative is to propagate the entire useful information to reconstruct the pruned layer instead of directly discarding the less important neurons. To this end, we propose a novel Layer DecompositionRecomposition Framework (LDRF) for neuron pruning, by which each layer’s output information is recovered in an embedding space and then propagated to reconstruct the following pruned layers with useful information preserved. We mainly conduct our experiments on ILSVRC-12 benchmark with VGG-16 and ResNet-50. What should be emphasized is that our results before end-to-end fine-tuning are significantly superior owing to the information-preserving property of our proposed framework. With end-to-end fine-tuning, we achieve state-of-the-art results of 5.13× and 3× speed-up with only 0.5% and 0.65% top-5 accuracy drop respectively, which outperform the existing neuron pruning methods.

Download Full-text

Fully Convolutional Networks and Geographic Object-Based Image Analysis for the Classification of VHR Imagery

Remote Sensing ◽

10.3390/rs11050597 ◽

2019 ◽

Vol 11 (5) ◽

pp. 597 ◽

Cited By ~ 21

Author(s):

Nicholus Mboga ◽

Stefanos Georganos ◽

Tais Grippa ◽

Moritz Lennert ◽

Sabine Vanhuysse ◽

...

Keyword(s):

Democratic Republic Of Congo ◽

State Of The Art ◽

Computational Cost ◽

Convolutional Networks ◽

Object Based Image Analysis ◽

Fully Convolutional Networks ◽

Object Based ◽

Geographic Object ◽

End To End ◽

Future Work

Land cover Classified maps obtained from deep learning methods such as Convolutional neural networks (CNNs) and fully convolutional networks (FCNs) usually have high classification accuracy but with the detailed structures of objects lost or smoothed. In this work, we develop a methodology based on fully convolutional networks (FCN) that is trained in an end-to-end fashion using aerial RGB images only as input. Skip connections are introduced into the FCN architecture to recover high spatial details from the lower convolutional layers. The experiments are conducted on the city of Goma in the Democratic Republic of Congo. We compare the results to a state-of-the art approach based on a semi-automatic Geographic object image-based analysis (GEOBIA) processing chain. State-of-the art classification accuracies are obtained by both methods whereby FCN and the best baseline method have an overall accuracy of 91.3% and 89.5% respectively. The maps have good visual quality and the use of an FCN skip architecture minimizes the rounded edges that is characteristic of FCN maps. Additional experiments are done to refine FCN classified maps using segments obtained from GEOBIA generated at different scale and minimum segment size. High OA of up to 91.5% is achieved accompanied with an improved edge delineation in the FCN maps, and future work will involve explicitly incorporating boundary information from the GEOBIA segmentation into the FCN pipeline in an end-to-end fashion. Finally, we observe that FCN has a lower computational cost than the standard patch-based CNN approach especially at inference.

Download Full-text

Runtime mechanisms to survive new HPC architectures: A use case in human respiratory simulations

The International Journal of High Performance Computing Applications ◽

10.1177/1094342019842919 ◽

2019 ◽

Vol 34 (1) ◽

pp. 42-56 ◽

Cited By ~ 7

Author(s):

Marta Garcia-Gasulla ◽

Filippo Mantovani ◽

Marc Josep-Fabrego ◽

Beatriz Eguzkitza ◽

Guillaume Houzeaux

Keyword(s):

Particle Tracking ◽

High Performance ◽

State Of The Art ◽

Particle Dynamics ◽

Use Case ◽

Drug Effectiveness ◽

Performance Improvements ◽

Load Imbalance ◽

Computational Resources ◽

Performance Computing

Computational fluid and particle dynamics (CFPD) simulations are of paramount importance for studying and improving drug effectiveness. Computational requirements of CFPD codes demand high-performance computing (HPC) resources. For these reasons, we introduce and evaluate in this article system software techniques for improving performance and tolerating load imbalance on a state-of-the-art production CFPD code. We demonstrate benefits of these techniques on Intel-, IBM- and Arm-based HPC technologies ranked in the Top500 supercomputers, showing the importance of using mechanisms applied at runtime to improve the performance independently of the underlying architecture. We run a real CFPD simulation of particle tracking on the human respiratory system, showing performance improvements of up to 2×, across different architectures, while applying runtime techniques and keeping constant the computational resources.

Download Full-text