Greedy auto-augmentation for n-shot learning using deep neural networks

Naghizadeh, Alireza; Metaxas, Dimitris N.; Liu, Dongfang

doi:10.1016/j.neunet.2020.11.015

Cited by 15 publications

(17 citation statements)

References 18 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It is an open issue to perform a detailed study of how much the choice of different backgrounds (in a given set) affects the classifier performance. We also observe that the approach of [ 25 , 26 , 27 , 28 ] to automatically derive augmentation technique from data seems suitable to be applied also in the framework considered here.…”

Section: Related Workmentioning

confidence: 77%

“…Given a set of operators, finding an effective data-augmentation technique based on suitable compositions of such operators may greatly improve the overall classifier predictions. Although in many cases the proposed augmentation techniques are dataset-dependent, several techniques have also been proposed in the literature to learn an effective augmentation technique from the dataset itself, by searching a space of possible augmentation procedures [ 25 , 26 , 27 , 28 ]. In this paper, however, we concentrate on the case of data augmentation with respect to a different characteristic, i.e., image background.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Data Augmentation Using Background Replacement for Automated Sorting of Littered Waste

2021

View full text Add to dashboard Cite

The introduction of sophisticated waste treatment plants is making the process of trash sorting and recycling more and more effective and eco-friendly. Studies on Automated Waste Sorting (AWS) are greatly contributing to making the whole recycling process more efficient. However, a relevant issue, which remains unsolved, is how to deal with the large amount of waste that is littered in the environment instead of being collected properly. In this paper, we introduce BackRep: a method for building waste recognizers that can be used for identifying and sorting littered waste directly where it is found. BackRep consists of a data-augmentation procedure, which expands existing datasets by cropping solid waste in images taken on a uniform (white) background and superimposing it on more realistic backgrounds. For our purpose, realistic backgrounds are those representing places where solid waste is usually littered. To experiment with our data-augmentation procedure, we produced a new dataset in realistic settings. We observed that waste recognizers trained on augmented data actually outperform those trained on existing datasets. Hence, our data-augmentation procedure seems a viable approach to support the development of waste recognizers for urban and wild environments.

show abstract

Section: Related Workmentioning

confidence: 77%

Section: Related Workmentioning

confidence: 99%

Data Augmentation Using Background Replacement for Automated Sorting of Littered Waste

2021

View full text Add to dashboard Cite

show abstract

“…Step 2: Stochastic Gradient Descent for Meta-Model Training Initialize: Randomly initialize the parameters of [23].…”

Section: Statementmentioning

confidence: 99%

Ensemble-based Hybrid Optimization of Bayesian Neural Networks and Traditional Machine Learning Algorithms

Tan

2023

Preprint

View full text Add to dashboard Cite

This research introduces a novel methodology for optimizing Bayesian Neural Networks (BNNs) by synergistically integrating them with traditional machine learning algorithms such as Random Forests (RF), Gradient Boosting (GB), and Support Vector Machines (SVM). Utilizing ensemble methods, represented by the equation yensemble = M ∈M wM · yM , and stacking techniques, the study formulates a unique hybrid predictive system. The research rigorously explores the properties of individual non-Bayesian models, establishing their feature importance , generalization error, and optimization landscapes through lemmas and theorems. It proves the optimality of the proposed ensemble method and the robustness of the stacking technique. Feature integration is mathematically formulated to achieve significant information gain. Additionally, in synthesizing the findings, our research corroborates the mathematical formulations underlying ensemble methods while offering nuanced insights into the limitations of hyperparameter tuning. Specifically, the ensemble method empirically validates the ensemble generalization error equation Eensemble = n i=1 w 2 i ϵi + 2 n i=1 j̸ =i wiwjρ(Mi, Mj)ϵiϵj, showcasing the ensemble’s minimized generalization error. This is further optimized through the Lagrangian function L(w1, w2,. .. , wn, λ) = Eensemble + λ 1 − n i=1 wi , allowing for adaptive weight adjustments. Feature integration solidifies these results by emphasizing the second-order conditions for optimality, including stationarity (∇L = 0) and positive definiteness of the Hessian matrix. Conversely, hyperparameter tuning indicates a subdued impact in improving Expected Improvement (EI), represented by EI(x) = E[max(f (x) − f (x *), 0)]. Overall, the ensemble method stands out as a robust, algorithmically optimized approach.

show abstract

“…While big data would allow for training, data scientists may apply newer techniques with fewer data points to mine and transfer them [ 48 ], despite training on limited labeled information in the data [ 49 , 50 ]. Models for ML can be trained with small datasets using few-shot and n-shot approaches [ 51 , 52 ]. Few-shot learning has the potential to help clean and label datasets, as well as generate more data.…”

Section: Data Volumementioning

confidence: 99%

Enhanced Patient-Centricity: How the Biopharmaceutical Industry Is Optimizing Patient Care through AI/ML/DL

Zou

2022

Healthcare

View full text Add to dashboard Cite

Technologies utilizing cutting-edge methodologies, including artificial intelligence (AI), machine learning (ML) and deep learning (DL), present powerful opportunities to help evaluate, predict, and improve patient outcomes by drawing insights from real-world data (RWD) generated during medical care. They played a role during and following the Coronavirus Disease 2019 (COVID-19) pandemic by helping protect healthcare providers, prioritize care for vulnerable populations, predict disease trends, and find optimal therapies. Potential applications across therapeutic areas include diagnosis, disease management and patient journey mapping. Use of fit-for-purpose datasets for ML models is seeing growth and may potentially help additional enterprises develop AI strategies. However, biopharmaceutical companies often face specific challenges, including multi-setting data, system interoperability, data governance, and patient privacy requirements. There remains a need for evolving regulatory frameworks, operating models, and data governance to enable further developments and additional research. We explore recent literature and examine the hurdles faced by researchers in the biopharmaceutical industry to fully realize the promise of AI/ML/DL for patient-centric purposes.

show abstract

Greedy auto-augmentation for n-shot learning using deep neural networks

Cited by 15 publications

References 18 publications

Data Augmentation Using Background Replacement for Automated Sorting of Littered Waste

Data Augmentation Using Background Replacement for Automated Sorting of Littered Waste

Ensemble-based Hybrid Optimization of Bayesian Neural Networks and Traditional Machine Learning Algorithms

Enhanced Patient-Centricity: How the Biopharmaceutical Industry Is Optimizing Patient Care through AI/ML/DL

Contact Info

Product

Resources

About