Published Sep 6, 2021

MiniRocket

Explore how MiniRocket revolutionizes time series classification with a deterministic approach that enhances Rocket's speed by 75x, as Kyle Polich and Angus Dempster discuss its groundbreaking impact, cross-domain potential, and practical recommendations for implementation.

Episode Highlights

Topics covered

Episode Highlights

Related Episodes

Time Series Mini Episodes
Answers 383 questions
[MINI] Random Forest
Answers 383 questions
Ultra Long Time Series
Answers 383 questions
Graph Transformations
Answers 383 questions
[MINI] Automated Feature Engineering
Answers 383 questions
[MINI] Exponential Time Algorithms
Answers 383 questions
Graphs and ML for Robotics
Answers 383 questions
[MINI] AdaBoost
Answers 383 questions
Boosted Embeddings for Time Series
Answers 383 questions
[MINI] Auto-correlative functions and correlograms
Answers 383 questions
[MINI] k-means clustering
Answers 383 questions
[MINI] Logistic Regression on Audio Data
Answers 383 questions
Long Term Time Series Forecasting
Answers 383 questions
[MINI] Structured and Unstructured Data
Answers 383 questions
[MINI] MapReduce
Answers 383 questions

Dexa/Kyle Polich

MiniRocket

Topics covered

Popular Clips

Optimization Techniques

Scalability Challenges

Advancements in Time Series Approaches

Rocket: Transforming Time Series

Mini Rocket: Non-random Method

Exceptional Opportunities

Future of Time Series

Mini Rocket's Speed and Accuracy

Customer Insights

Benchmarking Time Series Classification

Time Series Classification Libraries

Episode Highlights

MiniRocket Overview

Future Directions

Time Series Classification

Related Episodes

Time Series Mini Episodes

[MINI] Random Forest

Ultra Long Time Series

Graph Transformations

[MINI] Automated Feature Engineering

[MINI] Exponential Time Algorithms

Graphs and ML for Robotics

[MINI] AdaBoost

Boosted Embeddings for Time Series

[MINI] Auto-correlative functions and correlograms

[MINI] k-means clustering

[MINI] Logistic Regression on Audio Data

Long Term Time Series Forecasting

[MINI] Structured and Unstructured Data

[MINI] MapReduce

MiniRocket

Topics covered

Popular Clips

Episode Highlights

MiniRocket Overview

Future DirectionsThe exploration of cross-domain influences in time series classification reveals the potential of adapting methods from image classification and NLP. Kyle Polich and Angus Dempster discuss the implementation of MiniRocket and recommend libraries for those interested in this field.

Future Directions

Time Series ClassificationAngus Dempster, a PhD student at Monash University, discusses the transformative impact of MiniRocket on time series classification. He explains how this innovation achieves a 75x speed improvement on larger datasets while maintaining performance.

Time Series Classification

Related Episodes