Learning Future Classifiers without Additional Data

Kumagai, Atsutoshi; Iwata, Tomoharu

doi:10.1609/aaai.v30i1.10208

Cited by 6 publications

(8 citation statements)

References 24 publications

(25 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Present is also considered as a weighting method, where there is a weight only at the current time. The AAAI16 method is for learning future classifiers by using the one-step algorithm (Kumagai and Iwata 2016). With the proposed method, we ran the learning four times with different initial conditions, and selected the result with the highest lower bound.…”

Section: Comparison Methodsmentioning

confidence: 99%

“…For Batch, Online and Present, we chose the regularization parameter from {10 −1 , 1, 10 1 } in terms of which average AUC over all test time units was the best. For AAAI16, we set the parameters for the gamma priors as the same values as those in a previous study (Kumagai and Iwata 2016), ran the learning four times with different initial conditions, and changed the order of vector autoregressive models from 1, . .…”

Section: Comparison Methodsmentioning

confidence: 99%

“…Although this method predicts the future state of a probability distribution, the proposed method predicts the future decision boundary. A method for learning future decision boundaries has also been proposed (Kumagai and Iwata 2016). Although it is assumed that the decision boundary depends linearly on the previous decision boundaries with this method, the proposed method does not rely on the such assumption due to GP regression, which is a non-linear non-parametric regression model.…”

Section: Related Workmentioning

confidence: 99%

“…To find the temporal variation of the classification performance in our experiments, we evaluated Area Under the Curve (AUC) by using data until a certain time unit for training, and the remaining for testing. This setting is the same as that in a previous study (Kumagai and Iwata 2016). For 0.988±0.002 0.940±0.011 0.443±0.036 0.860±0.020 0.860±0.020 Non-linear 0.984±0.003 0.952±0.012 0.302±0.031 0.851±0.021 0.854±0.020 SPAM2 0.956±0.003 0.907±0.006 0.930±0.005 0.928±0.004 0.915±0.005 ELEC2 0.921±0.005 0.926±0.005 0.916±0.004 0.904±0.005 0.903±0.005 ONP 0.705±0.002 0.533±0.004 0.630±0.002 0.622±0.002 0.622±0.002 BLOG 0.838±0.003 0.827±0.003 0.827±0.004 0.730±0.004 0.754±0.004 Linear, we set t T = 10 and the remaining ten time units as test data.…”

Section: Settingmentioning

confidence: 99%

“…To overcome this problem, a method for learning future decision boundaries given only labeled data with timestamps collected until the current time has been proposed (Kumagai and Iwata 2016). This method learns the dynamics of the decision boundary without additional data and predicts the future decision boundary.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Learning Non-Linear Dynamics of Decision Boundaries for Maintaining Classification Performance

Kumagai

Iwata

2017

AAAI

Self Cite

View full text Add to dashboard Cite

We propose a method that involves a probabilistic model for learning future classifiers for tasks in which decision boundaries nonlinearly change over time. In certain applications, such as spam-mail classification, the decision boundary dynamically changes over time. Accordingly, the performance of the classifiers will deteriorate quickly unless the classifiers are updated using additional data. However, collecting such data can be expensive or impossible. The proposed model alleviates this deterioration in performance without additional data by modeling the non-linear dynamics of the decision boundary using Gaussian processes. The method also involves our developed learning algorithm for our model based on empirical variational Bayesian inference by which uncertainty of dynamics can be incorporated for future classification. The effectiveness of the proposed method was demonstrated through experiments using synthetic and real-world data sets.

show abstract

Section: Comparison Methodsmentioning

confidence: 99%

Section: Comparison Methodsmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Section: Settingmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Learning Non-Linear Dynamics of Decision Boundaries for Maintaining Classification Performance

Kumagai

Iwata

2017

AAAI

Self Cite

View full text Add to dashboard Cite

show abstract

Just in Time: Personal Temporal Insights for Altering Model Decisions

Boer

Deutch

Frost

et al. 2019

2019 IEEE 35th International Conference on Data Engineering (ICDE)

View full text Add to dashboard Cite

The interpretability of complex Machine Learning models is coming to be a critical social concern, as they are increasingly used in human-related decision-making processes such as resume filtering or loan applications. Individuals receiving an undesired classification are likely to call for an explanation-preferably one that specifies what they should do in order to alter that decision when they reapply in the future. Existing work focuses on a single ML model and a single point in time, whereas in practice, both models and data evolve over time: an explanation for an application rejection in 2018 may be irrelevant in 2019 since in the meantime both the model and the applicant's data can change. To this end, we propose a novel framework that provides users with insights and plans for changing their classification in particular future time points. The solution is based on combining state-of-the-art algorithms for (single) model explanations, ones for predicting future models, and databasestyle querying of the obtained explanations. We propose to demonstrate the usefulness of our solution in the context of loan applications, and interactively engage the audience in computing and viewing suggestions tailored for applicants based on their unique characteristic.

show abstract

Foresee What You Will Learn: Data Augmentation for Domain Generalization in Non-stationary Environment

Zeng

Wang

Zhang

et al. 2023

AAAI

View full text Add to dashboard Cite

Existing domain generalization aims to learn a generalizable model to perform well even on unseen domains. For many real-world machine learning applications, the data distribution often shifts gradually along domain indices. For example, a self-driving car with a vision system drives from dawn to dusk, with the sky gradually darkening. Therefore, the system must be able to adapt to changes in ambient illuminations and continue to drive safely on the road. In this paper, we formulate such problems as Evolving Domain Generalization, where a model aims to generalize well on a target domain by discovering and leveraging the evolving pattern of the environment. We then propose Directional Domain Augmentation (DDA), which simulates the unseen target features by mapping source data as augmentations through a domain transformer. Specifically, we formulate DDA as a bi-level optimization problem and solve it through a novel meta-learning approach in the representation space. We evaluate the proposed method on both synthetic datasets and real-world datasets, and empirical results show that our approach can outperform other existing methods.

show abstract

Learning Future Classifiers without Additional Data

Cited by 6 publications

References 24 publications

Learning Non-Linear Dynamics of Decision Boundaries for Maintaining Classification Performance

Learning Non-Linear Dynamics of Decision Boundaries for Maintaining Classification Performance

Just in Time: Personal Temporal Insights for Altering Model Decisions

Foresee What You Will Learn: Data Augmentation for Domain Generalization in Non-stationary Environment

Contact Info

Product

Resources

About