Region-Based Global Reasoning Networks

Wang, Chuanming; Fu, Huiyuan; Ling, Charles X.; Du, Peilun; Ma, Huadóng

doi:10.1609/aaai.v34i07.6893

Cited by 6 publications

(5 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Our empirical demonstration showcases several common findings when working with cost-sensitive prediction tasks: as presented in the introduction, benchmark studies have repeatedly shown that no CSL method is always superior on all data sets and with all ML algorithms (Coussement, 2013; Seiffert et al, 2008; Sheng & Ling, 2006). At the same time, no ML algorithm is always superior.…”

Section: Discussionsupporting

confidence: 64%

“…Sheng and Ling (2006) developed a method that also relies on posterior probability estimates, but instead of the exact estimate only the correct order of probabilities is required. This method is called empirical thresholding (ET).…”

Section: In Psychological Sciencementioning

confidence: 99%

“…SMOTE achieved a similar performance, especially at lower cost ratios. Finally, Sheng and Ling (2006) conducted a benchmark experiment and came to the conclusion that ET and TW are on par with each other, both outperforming MetaCost. Additionally, ET was the more robust method for high-cost ratios.…”

Section: In Psychological Sciencementioning

confidence: 99%

See 2 more Smart Citations

Everything has its price: Foundations of cost-sensitive machine learning and its application in psychology.

Sterner

Goretzko

Pargent

2023

Psychological Methods

View full text Add to dashboard Cite

Psychology has seen an increase in the use of machine learning (ML) methods. In many applications, observations are classified into one of two groups (binary classification). Off-the-shelf classification algorithms assume that the costs of a misclassification (false positive or false negative) are equal. Because this is often not reasonable (e.g., in clinical psychology), cost-sensitive machine learning (CSL) methods can take different cost ratios into account. We present the mathematical foundations and introduce a taxonomy of the most commonly used CSL methods, before demonstrating their application and usefulness on psychological data, that is, the drug consumption data set (N = 1, 885) from the University of California Irvine ML Repository. In our example, all demonstrated CSL methods noticeably reduced mean misclassification costs compared to regular ML algorithms. We discuss the necessity for researchers to perform small benchmarks of CSL methods for their own practical application. Thus, our open materials provide R code, demonstrating how CSL methods can be applied within the mlr3 framework (https://osf.io/cvks7/).

show abstract

Section: Discussionsupporting

confidence: 64%

Section: In Psychological Sciencementioning

confidence: 99%

See 1 more Smart Citation

Everything has its price: Foundations of cost-sensitive machine learning and its application in psychology.

Sterner

Goretzko

Pargent

2023

Psychological Methods

View full text Add to dashboard Cite

show abstract

“…Achieving good predictive performance is challenging in these cases, as algorithms tend to predict the label of the larger classes too frequently. One strategy to address this problem is threshold tuning (Sheng & Ling, 2006). Usually, an observation is classified as belonging to the positive class in a binary classification task if the estimated probability for this observation is greater than 0.5.…”

Section: Discussionmentioning

confidence: 99%

Predictive Modeling With Psychological Panel Data

Pargent

Gönna

2018

Zeitschrift für Psychologie

View full text Add to dashboard Cite

Abstract. Longitudinal panels include several thousand participants and variables. Traditionally, psychologists analyze only a few variables – partly because common unregularized linear models perform poorly when the number of variables ( p) approaches the number of observations ( N). Predictive modeling methods can be used when N [Formula: see text] p situations arise in psychological research. We illustrate these techniques on exemplary variables from the German GESIS Panel, while describing the choice of preprocessing, model classes, resampling techniques, hyperparameter tuning, and performance measures. In analyses with about 2,000 subjects and variables each, we predict panelists’ gender, sick days, an evaluation of US President Trump, income, life satisfaction, and sleep satisfaction. Elastic net and random forest models were compared to dummy predictions in benchmark experiments. While good performance was achieved, the linear elastic net performed similar to the nonlinear random forest. Elastic nets were refitted to extract the ten most important predictors. Their interpretation validates our approach, and further modeling options are discussed. Code can be found at https://osf.io/zpse3/

show abstract

“…Different from skeleton-based action recognition where the skeleton data can be naturally seen as graph structure, generic action recognition methods employ GCNs to model the relations between fixed regions or objects. For example, [4,23,44] adopts GCN to build a reasoning module to model the relations between disjoint and distant regions. [21,48] takes dense object proposals as graph nodes and learns the relations between them.…”

Section: Related Workmentioning

confidence: 99%

Representing Videos as Discriminative Sub-graphs for Action Recognition

Liu¹,

Qiu²,

Pan³

et al. 2022

Preprint

View full text Add to dashboard Cite

Human actions are typically of combinatorial structures or patterns, i.e., subjects, objects, plus spatio-temporal interactions in between. Discovering such structures is therefore a rewarding way to reason about the dynamics of interactions and recognize the actions. In this paper, we introduce a new design of sub-graphs to represent and encode the discriminative patterns of each action in the videos. Specifically, we present MUlti-scale Sub-graph LEarning (MUSLE) framework that novelly builds space-time graphs and clusters the graphs into compact sub-graphs on each scale with respect to the number of nodes. Technically, MUSLE produces 3D bounding boxes, i.e., tubelets, in each video clip, as graph nodes and takes dense connectivity as graph edges between tubelets. For each action category, we execute online clustering to decompose the graph into sub-graphs on each scale through learning Gaussian Mixture Layer and select the discriminative sub-graphs as action prototypes for recognition. Extensive experiments are conducted on both Something-Something V1 & V2 and Kinetics-400 datasets, and superior results are reported when comparing to state-of-the-art methods. More remarkably, our MUSLE achieves to-date the best reported accuracy of 65.0% on Something-Something V2 validation set.

show abstract

Region-Based Global Reasoning Networks

Cited by 6 publications

References 15 publications

Everything has its price: Foundations of cost-sensitive machine learning and its application in psychology.

Everything has its price: Foundations of cost-sensitive machine learning and its application in psychology.

Predictive Modeling With Psychological Panel Data

Representing Videos as Discriminative Sub-graphs for Action Recognition

Contact Info

Product

Resources

About