Filters








29,708 Hits in 4.6 sec

Improving active learning recall via disjunctive boolean constraints

Emre Velipasaoglu, Hinrich Schütze, Jan O. Pedersen
2007 Proceedings of the 30th annual international ACM SIGIR conference on Research and development in information retrieval - SIGIR '07  
In this paper, we propose a method to increase active learning recall by constraining sampling to a document subset rich in relevant examples.  ...  Active learning efficiently hones in on the decision boundary between relevant and irrelevant documents, but in the process can miss entire clusters of relevant documents, yielding classifiers with low  ...  The first 50 iterations were ordinary AL (uncertainty sampling). After that, we constrained AL as explained above.  ... 
doi:10.1145/1277741.1277962 dblp:conf/sigir/VelipasaogluSP07 fatcat:k7exbrgqhvattkoz2g6p3tnwfe

Active Clustering with Model-Based Uncertainty Reduction

Caiming Xiong, David M. Johnson, Jason J. Corso
2017 IEEE Transactions on Pattern Analysis and Machine Intelligence  
Here, we propose a novel online framework for active semi-supervised spectral clustering that selects pairwise constraints as clustering proceeds, based on the principle of uncertainty reduction.  ...  The resulting model is used to estimate the uncertainty reduction potential of each sample in the dataset.  ...  ACTIVE CLUSTERING FRAMEWORK WITH CERTAIN-SAMPLE SETS Recall that "certain-sample sets" are sets such that any two samples in the same certain-sample set are constrained to reside in the same cluster, and  ... 
doi:10.1109/tpami.2016.2539965 pmid:26978555 fatcat:yvrb433525hipjmsy6nqx4qsdm

Active Clustering with Model-Based Uncertainty Reduction [article]

Caiming Xiong, David Johnson, Jason J. Corso
2014 arXiv   pre-print
Here, we propose a novel online framework for active semi-supervised spectral clustering that selects pairwise constraints as clustering proceeds, based on the principle of uncertainty reduction.  ...  The resulting model is used to estimate the uncertainty reduction potential of each sample in the dataset.  ...  ACTIVE CLUSTERING FRAMEWORK WITH CERTAIN-SAMPLE SETS Recall that "certain-sample sets" are sets such that any two samples in the same certain-sample set are constrained to reside in the same cluster, and  ... 
arXiv:1402.1783v2 fatcat:4z7mypvwyjgl7lba6pm52qy4na

Towards Fewer Labels: Support Pair Active Learning for Person Re-identification [article]

Dapeng Jin, Minxian Li
2022 arXiv   pre-print
Afterwards, we introduce a constrained clustering algorithm to propagate the relationships of labeled support pairs to other unlabeled samples.  ...  Specifically, we firstly design a dual uncertainty selection strategy to iteratively discover support pairs and require human annotations.  ...  from labeled support pairs also spread implicitly via the constrained clustering described in Sec.3.2.  ... 
arXiv:2204.10008v1 fatcat:r7mt4ijrcjd7dgbbztruwnjxjq

A Lagrangian Duality Approach to Active Learning [article]

Juan Elenter, Navid NaderiAlizadeh, Alejandro Ribeiro
2022 arXiv   pre-print
Our approach, which we refer to as Active Learning via Lagrangian dualitY, or ALLY, leverages this fact to select a diverse set of unlabeled samples with the highest estimated dual variables as our query  ...  We formulate the learning problem using constrained optimization, where each constraint bounds the performance of the model on labeled samples.  ...  Our proposed batch active learning approach, which we refer to as Active Learning via Lagrangian dualitY, or ALLY, selects a diverse batch by clustering the unlabeled samples in the embedding space using  ... 
arXiv:2202.04108v1 fatcat:6iqthbyufzelzljk2ek72oicqy

Accelerated Learning-Based Interactive Image Segmentation Using Pairwise Constraints

Jamshid Sourati, Deniz Erdogmus, Jennifer G. Dy, Dana H. Brooks
2014 IEEE Transactions on Image Processing  
Our work in this paper is based on constrained spectral clustering that iteratively incorporates user feedback by propagating it through the calculated affinities [17] .  ...  Here we employ active learning of optimal queries to guide user interaction.  ...  The first three applies active learning strategies based on three simulated users (cases (i), (ii) and (iii)) and the segmentation is learned via our constrained spectral clustering approach.  ... 
doi:10.1109/tip.2014.2325783 pmid:24860031 pmcid:PMC4096329 fatcat:2xjn3c3ry5bk7gyabs47rkoao4

Random-Fuzzy Chance-Constrained Programming Optimal Power Flow of Wind Integrated Power Considering Voltage Stability

Rui Ma, Xuan Li, Weicheng Gao, Peng Lu, Tieqiang Wang
2020 IEEE Access  
and susceptance of branch between bus i and j ;  is cluster of all buses connected with bus i.  ...  Under the DOPF schemes in the first and last iteration respectively and with random-fuzzy power injection samples of WG, the corresponding bus voltage sample distributions of bus 12 are as FIGURE 7. shows  ... 
doi:10.1109/access.2020.3040382 fatcat:73illdveoja5ti4dz4cpnnfxye

Active block diagonal subspace clustering

Ziqi Xie, Lihong Wang
2021 IEEE Access  
[38] proposed an uncertainty reduction model URASC (Uncertainty Reducing Active Spectral Clustering) for informative sample selection, and estimated the uncertainty reduction potential of each sample  ...  . • Rand: Randomly sample given number of unlabelled data points. • URASC [38] : An active spectral clustering algorithm based on uncertainty reduction. • URASC+AL: Unlabeled data points are selected  ... 
doi:10.1109/access.2021.3087575 fatcat:pkw4b2npdvb2zc3crj5yyk3iuy

Interactive Bayesian Hierarchical Clustering [article]

Sharad Vikram, Sanjoy Dasgupta
2016 arXiv   pre-print
We design an interactive Bayesian algorithm that incorporates user interaction into hierarchical clustering while still utilizing the geometry of the data by sampling a constrained posterior distribution  ...  To address this, several methods incorporate constraints obtained from users into clustering algorithms, but unfortunately do not apply to hierarchical clustering.  ...  (sketch) To show irreducibility, we show that a tree T has an non-zero probability of reaching an arbitrary tree T via constrained-SPR moves where both T and T satisfy a set of triplets C.  ... 
arXiv:1602.03258v3 fatcat:m5t7iksp3jdhrixgbrrw6yjtz4

Measurement of quasielastic-like neutrino scattering at ⟨Eν⟩∼3.5 GeV on a hydrocarbon target

D. Ruterbories, K. Hurtado, J. Osta, F. Akbar, L. Aliaga, D. A. Andrade, M. V. Ascencio, A. Bashyal, A. Bercellie, M. Betancourt, A. Bodek, H. Budd (+62 others)
2019 Physical Review D  
In order to reduce model dependence the backgrounds are constrained using independent data samples.  ...  regions for non-fluxintegrated results.  ... 
doi:10.1103/physrevd.99.012004 fatcat:z7f2tgxbnnarlfpqwyndjzyn6a

Active learning for interactive satellite image change detection [article]

Hichem Sahbi and Sebastien Deschamps and Andrei Stoian
2021 arXiv   pre-print
according to the oracle's responses, updates a decision function iteratively.  ...  We introduce in this paper a novel active learning algorithm for satellite image change detection.  ...  Uncertainty sampling consists in choosing the display whose unlabeled samples are the most ambiguous (i.e., whose SVM scores are the closest to zero).  ... 
arXiv:2110.04250v1 fatcat:xowd46f2knfzvmr4e6wdvgnpr4

OUP accepted manuscript

2019 Monthly notices of the Royal Astronomical Society  
The BAX clusters are tested for the presence of sub-structures, acting as proxies for core mergers, culminating in sub-samples of 8 merging and 25 non-merging galaxy clusters.  ...  We select our main cluster sample from the X-ray Galaxy Clusters Database (BAX), which are populated with Sloan Digital Sky Survey (SDSS) galaxies.  ...  Figs 11 and 12 illustrate the constrained sample v rot composites.  ... 
doi:10.1093/mnras/stz2927 fatcat:y2aw44nnajf4znqeivotpjsqay

Constraining the parameters of high-dimensional models with active learning

Sascha Caron, Tom Heskes, Sydney Otten, Bob Stienen
2019 European Physical Journal C: Particles and Fields  
This makes it possible to constrain model parameters more efficiently than is currently done with the most common sampling algorithms and to train better performing machine learning models on the same  ...  In this paper we show that this problem can be alleviated by the use of active learning.  ...  From this new data the points with the highest uncertainty in their labelling are selected for sampling via the true sampling procedure and added to the data set.  ... 
doi:10.1140/epjc/s10052-019-7437-5 fatcat:crcos34rsbaznbur45fx3capby

Improving the Applicability of AI for Psychiatric Applications through Human-in-the-loop Methodologies

Chelsea Chandler, Peter W Foltz, Brita Elvevåg
2022 Schizophrenia Bulletin  
more efficiently than classic random sampling would otherwise allow.  ...  Specifically, we applied active learning strategies to the automatic scoring of a story recall task and compared the results to a traditional approach.  ...  Each active learning iteration sampled 100 additional items: 45% via uncertainty sampling, 45% via diversity sampling, and 10% via random sampling.  ... 
doi:10.1093/schbul/sbac038 pmid:35639561 fatcat:xneskde6irhu3lmnoiq7xpkkxm

Exploration of Metamodeling Sampling Criteria for Constrained Global Optimization

Michael J. Sasena, Panos Papalambros, Pierre Goovaerts
2002 Engineering optimization (Print)  
Criteria that place more emphasis on global searching require more iterations to locate optima and do so less accurately than criteria emphasizing local search.  ...  The infill sampling criterion has a strong influence on how efficiently and accurately EGO locates the optimum.  ...  Typically, the kriging model parameters are fit at each iteration via Maximum Likelihood Estimation (MLE) [12] .  ... 
doi:10.1080/03052150211751 fatcat:opffxzlmnjetnkg7rnctzm5wwi
« Previous Showing results 1 — 15 out of 29,708 results