Tag Archives: Deep Neural Networks

Leveraging the unexplainability and opacity of NNs to generate random numbers

Y. Almardeny, A. Benavoli, N. Boujnah and E. Naredo, A Reinforcement Learning System for Generating Instantaneous Quality Random Sequences, IEEE Transactions on Artificial Intelligence, vol. 4, no. 3, pp. 402-415, June 2023 DOI: 10.1109/TAI.2022.3161893.

Random numbers are essential to most computer applications. Still, producing high-quality random sequences is a big challenge. Inspired by the success of artificial neural networks and reinforcement learning, we propose a novel and effective end-to-end learning system to generate pseudorandom sequences that operates under the upside-down reinforcement learning framework. It is based on manipulating the generalized information entropy metric to derive commands that instantaneously guide the agent toward the optimal random behavior. Using a wide range of evaluation tests, the proposed approach is compared against three state-of-the-art accredited pseudorandom number generators (PRNGs). The experimental results agree with our theoretical study and show that the proposed framework is a promising candidate for a wide range of applications.

Embedding actual knowledge into Deep Learning to improve its reliability

Lutter M, Peters J., Combining physics and deep learning to learn continuous-time dynamics models, The International Journal of Robotics Research. 2023;42(3):83-107 DOI: 10.1177/02783649231169492.

Deep learning has been widely used within learning algorithms for robotics. One disadvantage of deep networks is that these networks are black-box representations. Therefore, the learned approximations ignore the existing knowledge of physics or robotics. Especially for learning dynamics models, these black-box models are not desirable as the underlying principles are well understood and the standard deep networks can learn dynamics that violate these principles. To learn dynamics models with deep networks that guarantee physically plausible dynamics, we introduce physics-inspired deep networks that combine first principles from physics with deep learning. We incorporate Lagrangian mechanics within the model learning such that all approximated models adhere to the laws of physics and conserve energy. Deep Lagrangian Networks (DeLaN) parametrize the system energy using two networks. The parameters are obtained by minimizing the squared residual of the Euler\u2013Lagrange differential equation. Therefore, the resulting model does not require specific knowledge of the individual system, is interpretable, and can be used as a forward, inverse, and energy model. Previously these properties were only obtained when using system identification techniques that require knowledge of the kinematic structure. We apply DeLaN to learning dynamics models and apply these models to control simulated and physical rigid body systems. The results show that the proposed approach obtains dynamics models that can be applied to physical systems for real-time control. Compared to standard deep networks, the physics-inspired models learn better models and capture the underlying structure of the dynamics.

Using CNNs trained with image data to predict time series data

Aniello De Santo, Antonino Ferraro, Antonio Galli, Vincenzo Moscato, Giancarlo Sperl�, Evaluating time series encoding techniques for Predictive Maintenance, Expert Systems with Applications, Volume 210, 2022 DOI: 10.1016/j.eswa.2022.118435.

Predictive Maintenance has become an important component in modern industrial scenarios, as a way to minimize down-times and fault rate for different equipment. In this sense, while machine learning and deep learning approaches are promising due to their accurate predictive abilities, their data-heavy requirements make them significantly limited in real world applications. Since one of the main issues to overcome is lack of consistent training data, recent work has explored the possibility of adapting well-known deep-learning models for image recognition, by exploiting techniques to encode time series as images. In this paper, we propose a framework for evaluating some of the best known time series encoding techniques, together with Convolutional Neural Network-based image classifiers applied to predictive maintenance tasks. We conduct an extensive empirical evaluation of these approaches for the failure prediction task on two real-world datasets (PAKDD2020 Alibaba AI OPS Competition and NASA bearings), also comparing their performances with respect to the state-of-the-art approaches. We further discuss advantages and limitation of the exploited models when coupled with proper data augmentation techniques.

Reducing outliers in time series with singular spectrum analysis and use of deep learning for change detection

Muktesh Gupta, Rajesh Wadhvani, Akhtar Rasool, Real-time Change-Point Detection: A deep neural network-based adaptive approach for detecting changes in multivariate time series data, Expert Systems with Applications, Volume 209, 2022 DOI: 10.1016/j.eswa.2022.118260.

The behavior of a time series may be affected by various factors. Changes in mean, variance, frequency, and auto-correlation are the most common. Change-Point Detection (CPD) aims to track down abrupt statistical characteristic changes in time series that can benefit many applications in different domains. As demonstrated in recently introduced CPD methodologies, deep learning approaches have the potential to identify more subtle changes. However, due to improper handling of data and insufficient training, these methodologies generate more false alarms and are not efficient enough in detecting change-points. In real-time CPD algorithms, preprocessed data plays a vital role in increasing the algorithm\u2019s efficiency and minimizing false alarm rates. Therefore, preprocessing of data should be a part of the algorithm, but in the existing methods, preprocessing of data is done initially, and then the whole dataset is passed to the CPD algorithm. A new three-phase architecture is proposed to address this issue, in which all phases, from preprocessing to CPD, work in an adaptive manner. The phases are integrated into a pipeline, allowing the algorithm to work in real-time. Our proposed strategy performs optimally and consistently based on performance metrics resulting from experiments on real-world datasets and artifacts. This work effectively addresses the issue of non-stationary data normalization using deep learning approaches. To reduce noise and outliers from the data, a recursive version of singular spectrum analysis is introduced. It is demonstrated that the method\u2019s performance has significantly improved by combining adaptive preprocessing with deep learning CPD techniques.

NOTE: See also C. Ma, L. Zhang, W. Pedrycz and W. Lu, “The Long-Term Prediction of Time Series: A Granular Computing-Based Design Approach,” in IEEE Transactions on Systems, Man, and Cybernetics: Systems, vol. 52, no. 10, pp. 6326-6338, Oct. 2022, doi: 10.1109/TSMC.2022.3144395.

See also https://babel.isa.uma.es/kipr/?p=1548

Example of non-NN approach that produces better results in classification tasks than NNs

Jiang, Zhiying and Yang, Matthew and Tsirlin, Mikhail and Tang, Raphael and Dai, Yiqin and Lin, Jimmy, Low-Resource Text Classification: A Parameter-Free Classification Method with Compressors, . Findings of the Association for Computational Linguistics: ACL 2023 URL.

Deep neural networks (DNNs) are often used for text classification due to their high accuracy. However, DNNs can be computationally intensive, requiring millions of parameters and large amounts of labeled data, which can make them expensive to use, to optimize, and to transfer to out-of-distribution (OOD) cases in practice. In this paper, we propose a non-parametric alternative to DNNs that??s easy, lightweight, and universal in text classification: a combination of a simple compressor like gzip with a k-nearest-neighbor classifier. Without any training parameters, our method achieves results that are competitive with non-pretrained deep learning methods on six in-distribution datasets.It even outperforms BERT on all five OOD datasets, including four low-resource languages. Our method also excels in the few-shot setting, where labeled data are too scarce to train DNNs effectively.

Discrete Q-learning used, along a Deep CNN for localization, for mobile robot navigation

Amirhossein Shantia, Rik Timmers, Yiebo Chong, Cornel Kuiper, Francesco Bidoia, Lambert Schomaker, Marco Wiering, Two-stage visual navigation by deep neural networks and multi-goal reinforcement learning, . Robotics and Autonomous Systems, Volume 138, 2021 DOI: 10.1016/j.robot.2021.103731.

In this paper, we propose a two-stage learning framework for visual navigation in which the experience of the agent during exploration of one goal is shared to learn to navigate to other goals. We train a deep neural network for estimating the robot’s position in the environment using ground truth information provided by a classical localization and mapping approach. The second simpler multi-goal Q-function learns to traverse the environment by using the provided discretized map. Transfer learning is applied to the multi-goal Q-function from a maze structure to a 2D simulator and is finally deployed in a 3D simulator where the robot uses the estimated locations from the position estimator deep network. In the experiments, we first compare different architectures to select the best deep network for location estimation, and then compare the effects of the multi-goal reinforcement learning method to traditional reinforcement learning. The results show a significant improvement when multi-goal reinforcement learning is used. Furthermore, the results of the location estimator show that a deep network can learn and generalize in different environments using camera images with high accuracy in both position and orientation.

“Early exit” deep neural networks (i.e., CNN that provide outputs at intermediate points)

Scardapane, S., Scarpiniti, M., Baccarelli, E. et al. , Why Should We Add Early Exits to Neural Networks? . Cogn Comput 12, 954–966 (2020) DOI: 10.1007/s12559-020-09734-4.

Deep neural networks are generally designed as a stack of differentiable layers, in which a prediction is obtained only after running the full stack. Recently, some contributions have proposed techniques to endow the networks with early exits, allowing to obtain predictions at intermediate points of the stack. These multi-output networks have a number of advantages, including (i) significant reductions of the inference time, (ii) reduced tendency to overfitting and vanishing gradients, and (iii) capability of being distributed over multi-tier computation platforms. In addition, they connect to the wider themes of biological plausibility and layered cognitive reasoning. In this paper, we provide a comprehensive introduction to this family of neural networks, by describing in a unified fashion the way these architectures can be designed, trained, and actually deployed in time-constrained scenarios. We also describe in-depth their application scenarios in 5G and Fog computing environments, as long as some of the open research questions connected to them.

Simultaneous localization, mapping and semantic labelling in mobile robots

Taniguchi, Akira, Hagiwara, Yoshinobu, Taniguchi, Tadahiro, Inamura, Tetsunari, Improved and scalable online learning of spatial concepts and language models with mapping, Autonomous Robots 44(6), DOI: 10.1007/s10514-020-09905-0.

We propose a novel online learning algorithm, called SpCoSLAM 2.0, for spatial concepts and lexical acquisition with high accuracy and scalability. Previously, we proposed SpCoSLAM as an online learning algorithm based on unsupervised Bayesian probabilistic model that integrates multimodal place categorization, lexical acquisition, and SLAM. However, our original algorithm had limited estimation accuracy owing to the influence of the early stages of learning, and increased computational complexity with added training data. Therefore, we introduce techniques such as fixed-lag rejuvenation to reduce the calculation time while maintaining an accuracy higher than that of the original algorithm. The results show that, in terms of estimation accuracy, the proposed algorithm exceeds the original algorithm and is comparable to batch learning. In addition, the calculation time of the proposed algorithm does not depend on the amount of training data and becomes constant for each step of the scalable algorithm. Our approach will contribute to the realization of long-term spatial language interactions between humans and robots.

Application of Deep RL to person following by a robot, reducing the training effort of the network by reusing simple state situations in many artificially generated states

Pang, L., Zhang, Y., Coleman, S. et al., Efficient Hybrid-Supervised Deep Reinforcement Learning for Person Following Robot, J Intell Robot Syst 97, 299–312 (2020), DOI: 10.1007/s10846-019-01030-0.

Traditional person following robots usually need hand-crafted features and a well-designed controller to follow the assigned person. Normally it is difficult to be applied in outdoor situations due to variability and complexity of the environment. In this paper, we propose an approach in which an agent is trained by hybrid-supervised deep reinforcement learning (DRL) to perform a person following task in end-to-end manner. The approach enables the robot to learn features autonomously from monocular images and to enhance performance via robot-environment interaction. Experiments show that the proposed approach is adaptive to complex situations with significant illumination variation, object occlusion, target disappearance, pose change, and pedestrian interference. In order to speed up the training process to ensure easy application of DRL to real-world robotic follower controls, we apply an integration method through which the agent receives prior knowledge from a supervised learning (SL) policy network and reinforces its performance with a value-based or policy-based (including actor-critic method) DRL model. We also utilize an efficient data collection approach for supervised learning in the context of person following. Experimental results not only verify the robustness of the proposed DRL-based person following robot system, but also indicate how easily the robot can learn from mistakes and improve performance.

A nice (short) survey of deep RL

Matthew Botvinick, Sam Ritter, Jane X. Wang, Zeb Kurth-Nelson, Charles Blundell, Demis Hassabis, Reinforcement Learning, Fast and Slow, Trends in Cognitive Sciences, Volume 23, Issue 5, 2019, Pages 408-422 DOI: 10.1016/j.tics.2019.02.006.

Deep reinforcement learning (RL) methods have driven impressive advances in artificial intelligence in recent years, exceeding human performance in domains ranging from Atari to Go to no-limit poker. This progress has drawn the attention of cognitive scientists interested in understanding human learning. However, the concern has been raised that deep RL may be too sample-inefficient – that is, it may simply be too slow – to provide a plausible model of how humans learn. In the present review, we counter this critique by describing recently developed techniques that allow deep RL to operate more nimbly, solving problems much more quickly than previous methods. Although these techniques were developed in an AI context, we propose that they may have rich implications for psychology and neuroscience. A key insight, arising from these AI methods, concerns the fundamental connection between fast RL and slower, more incremental forms of learning.