Efficient Autotuning of Hyperparameters in Approximate Nearest Neighbor Search

Jääsaari, Elias; Hyvönen, Ville; Roos, Teemu

doi:10.1007/978-3-030-16145-3_46

Cited by 7 publications

(4 citation statements)

References 16 publications

(41 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Easy speed-ups include replacing the current NN method 53 by approximated and parallelized versions such as FLANN. 51,52 Multiple random projection trees (MRPT) 77,78 reduces expensive distance evaluations, thereby achieving higher speed than ANN and FLANN. However, MRPT has several issues: an insufficient number of requested nearest neighbors are returned; single-precision floating point is used; retrieving distances between neighbor points is not easily supported; and inaccurate search – in the worst case, points that are far away from the query point are returned as nearest neighbors.…”

Section: Discussionmentioning

confidence: 99%

Visual cluster separation using high-dimensional sharpened dimensionality reduction

Kim¹,

Telea

Trager

et al. 2022

Information Visualization

View full text Add to dashboard Cite

Applying dimensionality reduction (DR) to large, high-dimensional data sets can be challenging when distinguishing the underlying high-dimensional data clusters in a 2D projection for exploratory analysis. We address this problem by first sharpening the clusters in the original high-dimensional data prior to the DR step using Local Gradient Clustering (LGC). We then project the sharpened data from the high-dimensional space to 2D by a user-selected DR method. The sharpening step aids this method to preserve cluster separation in the resulting 2D projection. With our method, end-users can label each distinct cluster to further analyze an otherwise unlabeled data set. Our “High-Dimensional Sharpened DR” (HD-SDR) method, tested on both synthetic and real-world data sets, is favorable to DR methods with poor cluster separation and yields a better visual cluster separation than these DR methods with no sharpening. Our method achieves good quality (measured by quality metrics) and scales computationally well with large high-dimensional data. To illustrate its concrete applications, we further apply HD-SDR on a recent astronomical catalog.

show abstract

Section: Discussionmentioning

confidence: 99%

Visual cluster separation using high-dimensional sharpened dimensionality reduction

Kim¹,

Telea

Trager

et al. 2022

Information Visualization

View full text Add to dashboard Cite

show abstract

“…Hyperparameter tuning is an important optimization technique in machine learning algorithms, particularly for k-NN [38]. A grid search is essential for selecting the best hyperparameter configuration, which includes the value of 'k' and the wsix scheme used in k-NN [39].…”

Section: Hyperparameter Tuningmentioning

confidence: 99%

Improving ventilation classification in under-actuated zones: a k-nearest neighbor and data preprocessing approach

Yaddarabullah,

Rahman,

Saad

2024

IJEECS

View full text Add to dashboard Cite

This study investigates the use of k-nearest neighbors (k-NN) for classifying occupant positions in under-actuated zones, aiming to enhance ventilation control. The focus is on evaluating different data preprocessing techniques, particularly cumulative moving average (CMA), Kalman filtering (KF), and their combination, to boost the k-NN model's reliability and accuracy. The research uses received signal strength indicator (RSSI) data in a controlled setting. The methodology involves dividing the dataset into training and testing subsets and using root mean squared error (RMSE) to determine the best k value for model validation. The study performs a comparative analysis of the k-NN model's performance with both original and preprocessed RSSI data, focusing on metrics such as accuracy, precision, recall, F1-score, and RMSE. The findings emphasize the significant impact of the combined CMA-KF preprocessing technique in improving the model's accuracy and reliability. Specifically, this approach achieved an accuracy of 98.58%. The RMSE values are particularly noteworthy, exhibiting a perfect fit (RMSE of 0) for training data and a remarkably low RMSE of 0.119 for testing data, confirming the model's high accuracy and predictive capability.

show abstract

“…MRPT starts with creating a binary space-partitioning tree index, utilizing the auto-tuning process proposed in [29]. This process utilizes the sparse random-projections trees (sRPT) for which the parametrization is done according to the nearest neighbors' target recall level, provided to the algorithm along with the (k) number of nearest neighbors to search for.…”

Section: Speeding Up the Knn Classificationmentioning

confidence: 99%

Approximate kNN Classification for Biomedical Data

Anagnostou

Barbas

Vrahatis

et al. 2020

2020 IEEE International Conference on Big Data (Big Data)

View full text Add to dashboard Cite

We are in the era where the Big Data analytics has changed the way of interpreting the various biomedical phenomena, and as the generated data increase, the need for new machine learning methods to handle this evolution grows. An indicative example is the single-cell RNA-seq (scRNA-seq), an emerging DNA sequencing technology with promising capabilities but significant computational challenges due to the large-scaled generated data. Regarding the classification process for scRNAseq data, an appropriate method is the k Nearest Neighbor (kNN) classifier since it is usually utilized for large-scale prediction tasks due to its simplicity, minimal parameterization, and model-free nature. However, the ultra-high dimensionality that characterizes scRNA-seq impose a computational bottleneck, while prediction power can be affected by the "Curse of Dimensionality". In this work, we proposed the utilization of approximate nearest neighbor search algorithms for the task of kNN classification in scRNA-seq data focusing on a particular methodology tailored for high dimensional data. We argue that even relaxed approximate solutions will not affect the prediction performance significantly. The experimental results confirm the original assumption by offering the potential for broader applicability.

show abstract

Efficient Autotuning of Hyperparameters in Approximate Nearest Neighbor Search

Cited by 7 publications

References 16 publications

Visual cluster separation using high-dimensional sharpened dimensionality reduction

Visual cluster separation using high-dimensional sharpened dimensionality reduction

Improving ventilation classification in under-actuated zones: a k-nearest neighbor and data preprocessing approach

Approximate kNN Classification for Biomedical Data

Contact Info

Product

Resources

About