scholarly journals Blind Spot Detection for Safe Sim-to-Real Transfer

2020 ◽  
Vol 67 ◽  
pp. 191-234 ◽  
Author(s):  
Ramya Ramakrishnan ◽  
Ece Kamar ◽  
Debadeepta Dey ◽  
Eric Horvitz ◽  
Julie Shah
Keyword(s):  
Supervised Learning ◽  
Real World ◽  
A Priori ◽  
Class Imbalance ◽  
Predictive Performance ◽  
Blind Spot ◽  
Spot Detection ◽  
Real World Applications ◽  
Label Aggregation ◽  
Blind Spots

Agents trained in simulation may make errors when performing actions in the real world due to mismatches between training and execution environments. These mistakes can be dangerous and difficult for the agent to discover because the agent is unable to predict them a priori. In this work, we propose the use of oracle feedback to learn a predictive model of these blind spots in order to reduce costly errors in real-world applications. We focus on blind spots in reinforcement learning (RL) that occur due to incomplete state representation: when the agent lacks necessary features to represent the true state of the world, and thus cannot distinguish between numerous states. We formalize the problem of discovering blind spots in RL as a noisy supervised learning problem with class imbalance. Our system learns models for predicting blind spots within unseen regions of the state space by combining techniques for label aggregation, calibration, and supervised learning. These models take into consideration noise emerging from different forms of oracle feedback, including demonstrations and corrections. We evaluate our approach across two domains and demonstrate that it achieves higher predictive performance than baseline methods, and also that the learned model can be used to selectively query an oracle at execution time to prevent errors. We also empirically analyze the biases of various feedback types and how these biases influence the discovery of blind spots. Further, we include analyses of our approach that incorporate relaxed initial optimality assumptions. (Interestingly, relaxing the assumptions of an optimal oracle and an optimal simulator policy helped our models to perform better.) We also propose extensions to our method that are intended to improve performance when using corrections and demonstrations data.

An Empirical Study of Boosting Methods on Severely Imbalanced Data

2014 ◽  
Vol 513-517 ◽  
pp. 2510-2513 ◽  
Author(s):  
Xu Ying Liu
Keyword(s):  
Empirical Study ◽  
Real World ◽  
Class Imbalance ◽  
Imbalanced Data ◽  
Real World Applications ◽  
Under Sampling ◽  
The Difference ◽  
Imbalance Learning ◽  
Class Imbalance Learning ◽  
F Measure

Nowadays there are large volumes of data in real-world applications, which poses great challenge to class-imbalance learning: the large amount of the majority class examples and severe class-imbalance. Previous studies on class-imbalance learning mainly focused on relatively small or moderate class-imbalance. In this paper we conduct an empirical study to explore the difference between learning with small or moderate class-imbalance and learning with severe class-imbalance. The experimental results show that: (1) Traditional methods cannot handle severe class-imbalance effectively. (2) AUC, G-mean and F-measure can be very inconsistent for severe class-imbalance, which seldom appears when class-imbalance is moderate. And G-mean is not appropriate for severe class-imbalance learning because it is not sensitive to the change of imbalance ratio. (3) When AUC and G-mean are evaluation metrics, EasyEnsemble is the best method, followed by BalanceCascade and under-sampling. (4) A little under-full balance is better for under-sampling to handle severe class-imbalance. And it is important to handle false positives when design methods for severe class-imbalance.

scholarly journals Overcoming Blind Spots in the Real World: Leveraging Complementary Abilities for Joint Execution

2019 ◽  
Vol 33 ◽  
pp. 6137-6145
Author(s):  
Ramya Ramakrishnan ◽  
Ece Kamar ◽  
Besmira Nushi ◽  
Debadeepta Dey ◽  
Julie Shah ◽  
...  
Keyword(s):  
Real World ◽  
Cost Effective ◽  
Blind Spot ◽  
Imitation Learning ◽  
Second Step ◽  
The Real ◽  
Open World ◽  
Hand Off ◽  
Noisy Labels ◽  
Blind Spots

Simulators are being increasingly used to train agents before deploying them in real-world environments. While training in simulation provides a cost-effective way to learn, poorly modeled aspects of the simulator can lead to costly mistakes, or blind spots. While humans can help guide an agent towards identifying these error regions, humans themselves have blind spots and noise in execution. We study how learning about blind spots of both can be used to manage hand-off decisions when humans and agents jointly act in the real-world in which neither of them are trained or evaluated fully. The formulation assumes that agent blind spots result from representational limitations in the simulation world, which leads the agent to ignore important features that are relevant for acting in the open world. Our approach for blind spot discovery combines experiences collected in simulation with limited human demonstrations. The first step applies imitation learning to demonstration data to identify important features that the human is using but that the agent is missing. The second step uses noisy labels extracted from action mismatches between the agent and the human across simulation and demonstration data to train blind spot models. We show through experiments on two domains that our approach is able to learn a succinct representation that accurately captures blind spot regions and avoids dangerous errors in the real world through transfer of control between the agent and the human.

scholarly journals Co-GCN for Multi-View Semi-Supervised Learning

2020 ◽  
Vol 34 (04) ◽  
pp. 4691-4698
Author(s):  
Shu Li ◽  
Wen-Tao Li ◽  
Wei Wang
Keyword(s):  
Supervised Learning ◽  
Real World ◽  
State Of The Art ◽  
Data Sets ◽  
Real World Data ◽  
Convolutional Network ◽  
The Past ◽  
Real World Applications ◽  
Supervised Methods ◽  
Disjoint Sets

In many real-world applications, the data have several disjoint sets of features and each set is called as a view. Researchers have developed many multi-view learning methods in the past decade. In this paper, we bring Graph Convolutional Network (GCN) into multi-view learning and propose a novel multi-view semi-supervised learning method Co-GCN by adaptively exploiting the graph information from the multiple views with combined Laplacians. Experimental results on real-world data sets verify that Co-GCN can achieve better performance compared with state-of-the-art multi-view semi-supervised methods.

An Ensemble Framework of Multi-ratio Undersampling-based Imbalanced Classification

2021 ◽  
Vol 2 (1) ◽  
pp. 30-46
Author(s):  
Takahiro Komamizu ◽  
Yasuhiro Ogawa ◽  
Katsuhiko Toyama
Keyword(s):  
Real World ◽  
State Of The Art ◽  
Weighting Function ◽  
Class Imbalance ◽  
Classification Performance ◽  
Real World Data ◽  
Imbalanced Classification ◽  
Sampling Ratio ◽  
Real World Applications ◽  
F Measure

Class imbalance is commonly observed in real-world data, and it is problematic in that it degrades classification performance due to biased supervision. Undersampling is an effective resampling approach to the class imbalance. The conventional undersampling-based approaches involve a single fixed sampling ratio. However, different sampling ratios have different preferences toward classes. In this paper, an undersampling-based ensemble framework, MUEnsemble, is proposed. This framework involves weak classifiers of different sampling ratios, and it allows for a flexible design for weighting weak classifiers in different sampling ratios. To demonstrate the principle of the design, in this paper, a uniform weighting function and a Gaussian weighting function are presented. An extensive experimental evaluation shows that MUEnsemble outperforms undersampling-based and oversampling-based state-of-the-art methods in terms of recall, gmean, F-measure, and ROC-AUC metrics. Also, the evaluation showcases that the Gaussian weighting function is superior to the uniform weighting function. This indicates that the Gaussian weighting function can capture the different preferences of sampling ratios toward classes. An investigation into the effects of the parameters of the Gaussian weighting function shows that the parameters of this function can be chosen in terms of recall, which is preferred in many real-world applications.

Are Students Blind to Their Ethical Blind Spots? An Exploration of Why Ethics Education Should Focus on Self-Perception Biases

2017 ◽  
Vol 41 (4) ◽  
pp. 539-574 ◽  
Author(s):  
Kathleen A. Tomlin ◽  
Matthew L. Metzger ◽  
Jill Bradley-Geist ◽  
Tracy Gonzalez-Padron
Keyword(s):  
Real World ◽  
Ethics Education ◽  
Blind Spot ◽  
Ethical Challenges ◽  
Training Module ◽  
Self Perception ◽  
Behavioral Ethics ◽  
Bias Blind Spot ◽  
Psychology Literature ◽  
Blind Spots

Ethics blind spots, which have become a keystone of the emerging behavioral ethics literature, are essentially biases, heuristics, and psychological traps. Though students typically recognize that ethical challenges exist in the world at large, they often fail to see when they are personally prone to ethics blind spots. This creates an obstacle for ethics education—inducing students to act in an ethical manner when faced with real challenges. Grounded in the social psychology literature, we suggest that a meta-bias, the bias blind spot, should be addressed to facilitate student recognition of real-world ethical dilemmas and their own susceptibility to biases. We present a roadmap for an ethics education training module, developed to incorporate both ethics blind spots and self-perception biases. After completing the module, students identified potential ethical challenges in their real-world team projects and reflected on their susceptibility to ethical transgressions. Qualitative student feedback supports the value of this training module beyond traditional ethics education approaches. Lessons for management and ethics educators include (a) the value of timely, in-context ethics interventions and (b) the need for student self-reflection (more so than emphasis on broad ethical principles). Future directions are discussed.

Integrating Improved U-Net and Continuous Maximum Flow Algorithm for 3D Brain Tumor Image Segmentation

2020 ◽  
Vol 64 (4) ◽  
pp. 40412-1-40412-11
Author(s):  
Kexin Bai ◽  
Qiang Li ◽  
Ching-Hsin Wang
Keyword(s):  
Brain Tumor ◽  
Data Augmentation ◽  
A Priori ◽  
Class Imbalance ◽  
Maximum Flow ◽  
Magnetic Resonance Images ◽  
Tumor Segmentation ◽  
Similarity Coefficients ◽  
Segmentation Algorithms ◽  
Flow Algorithm

Abstract To address the issues of the relatively small size of brain tumor image datasets, severe class imbalance, and low precision in existing segmentation algorithms for brain tumor images, this study proposes a two-stage segmentation algorithm integrating convolutional neural networks (CNNs) and conventional methods. Four modalities of the original magnetic resonance images were first preprocessed separately. Next, preliminary segmentation was performed using an improved U-Net CNN containing deep monitoring, residual structures, dense connection structures, and dense skip connections. The authors adopted a multiclass Dice loss function to deal with class imbalance and successfully prevented overfitting using data augmentation. The preliminary segmentation results subsequently served as the a priori knowledge for a continuous maximum flow algorithm for fine segmentation of target edges. Experiments revealed that the mean Dice similarity coefficients of the proposed algorithm in whole tumor, tumor core, and enhancing tumor segmentation were 0.9072, 0.8578, and 0.7837, respectively. The proposed algorithm presents higher accuracy and better stability in comparison with some of the more advanced segmentation algorithms for brain tumor images.

Zur Medialität der Samenbank, oder: die Nacht der Substanz

2019 ◽  
Vol 10 (1) ◽  
pp. 39-62
Author(s):  
Nicole C. Karafyllis
Keyword(s):  
A Priori ◽  
Blind Spot ◽  
Sperm Bank ◽  
Cooling Chamber

Die moderne Samenbank lässt sich mit Medienbegriffen beschreiben, von Bestand bis Infrastruktur. Stets bleibt als blinder Fleck die Medialität des Samens, dessen Vitalität im Dunkel der Kühlkammer künstlich verlängert wird. Der Beitrag diskutiert Varianten, das Problem der Teleologie der Natur in Medienbegriffen abzuhandeln und bietet eine neue Geschichte von ›den Bienen und den Blumen‹. Er hebt den Samen als Inbegriff einer nicht reduzierbaren Substanz hervor, dessen Latenz als mediales Apriori des Lebenden begreifbar wird. A modern sperm bank can actually be described by using media terms such as ›stock‹ or ›infrastructure‹. However, the mediality of sperm seems to be persistently lingering in a blind spot, its vitality artificially prolonged in the dark of the cooling chamber. This article discusses different variants to treat the problem of describing the teleology of nature with the help of media terms and offers a new take on the story of ›the birds and the bees‹. The argumentation stresses the importance of sperm as the very quintessence of a non-reducible substance whose latency as a medial a priori of life thus becomes palpable

Sign in / Sign up

Export Citation Format

Share Document