Data Reduction Algorithms

An Overview of Key Algorithms in

Data Reduction
 Agenda

• Data Reduction: Overview

• Turning Point Algorithm
• AZTEC Algorithm
• Fan Algorithm
• Huffman Coding
• Modified Huffman Coding
• Run Length Coding
 Data Reduction Overview
Definition: Data reduction refers to the process of
minimizing the size of a dataset while preserving
its essential information.
Purpose: Reduces storage requirements and
transmission time.
Common Techniques: Lossless Compression
Lossy Compression
Applications:Image/Video Compression
Data Transmission, File Storage
 Turning Point Algorithm (TPA)
Overview: A technique for reducing data by
detecting and storing only significant data points
in a time series.
Steps: Identify the "turning points" (where the
slope changes significantly). Replace regions
between turning points with approximated
values.
Advantages: Efficient for time series data.
Preserves essential trends and patterns.
Applications: Time-series analysis in stock market
data, sensor data, etc.
 Mathematical Equation ( TPA)
.
 Algorithm Steps (TPA)
.
 Algorithm Steps (TPA)..
.
 Algorithm Steps (TPA)..
.
 Algorithm Steps (TPA)..
.
 AZTEC Algorithm
Adaptive Zero-Tracking and Error-Correction
• The AZTEC algorithm is a lossless compression technique
that is often used for compressing time-series data such as
ECG (Electrocardiogram) and EEG (Electroencephalogram)
signals. The algorithm combines differential encoding with
predictive encoding to reduce the amount of data that
needs to be stored or transmitted, while preserving the
integrity of the original signal.
• AZTEC, specifically, is efficient in handling continuous
signals with small variations between consecutive data
points, making it highly suitable for physiological signals like
ECG and EEG.
• The algorithm reduces the amount of data by storing only
the differences between adjacent values, rather than the
absolute values themselves.
 The AZTEC Algorithm: Overview
• AZTEC Algorithm: AZTEC stands for Adaptive Zero-
Tracking and Error-Correction.
• Primarily used for noise reduction and signal
enhancement.
• It is designed to improve the signal-to-noise ratio (SNR)
of biomedical signals.
• Main Features:Adaptive filtering to track noise or
artifacts in real-time.
• Error correction mechanisms to adjust the signal and
reduce interference.
• Suitable for a variety of biomedical applications, such
as EEG, ECG, and EMG.
 AZTEC Algorithm: Mathematical
Modeling
.
 AZTEC Algorithm: Mathematical
Modeling cont…
.
 Key Components of the AZTEC
Algorithm
• Adaptive Filtering: Uses an adaptive filter to model and remove
noise from the signal.
• The filter adjusts its parameters over time to adapt to changes in
noise characteristics.
• The goal is to minimize the error between the noisy signal and the
desired signal.
• Zero-Tracking: Focuses on tracking instances where the signal
crosses zero (important for detecting certain events in biomedical
signals like heartbeats or neural spikes).
• Helps in maintaining synchronization between the signal and
reference noise components.
• Error-Correction: Corrects distortions caused by noise or external
artifacts.
• Implements algorithms that adjust the signal dynamically, based on
the identified noise patterns.
 Fan Algorithm
Overview: The Fan algorithm is a method for
compressing data by encoding differences
between consecutive values rather than the raw
data.
Key Features: Data is grouped into
blocks,Differences within a block are encoded
using a smaller number of bits.
Advantages: Efficient for datasets with small
differences between consecutive values.
• Works well with data where redundancy is high.
Applications: Image compression, Video encoding
 Huffman Coding
Overview: Huffman coding is a popular lossless data
compression algorithm.
Principle: Assigns shorter codes to more frequent data
items and longer codes to less frequent items.
Steps:
– Calculate the frequency of each symbol.
– Build a binary tree based on frequencies (less frequent
symbols have longer codes).
– Assign binary codes based on tree structure.
Advantages:
– Optimal for static data with known frequencies.
Applications:
– Text file compression (e.g., ZIP files)
– Image compression (e.g., JPEG)
 Modified Huffman Coding
Overview: A variation of Huffman coding,
optimized for scenarios with multiple symbols
and complex datasets.
Differences from Standard Huffman:
– Adjusts for specific types of data and bit distribution.
– Can be used in conjunction with other compression
techniques.
Applications:
– Used in applications like fax compression (Group 3/4
FAX) and data communications.
 Run Length Coding (RLC)
Overview: A simple compression algorithm where
consecutive repeating characters are stored as a
single data value and a count.
Process:
– Identify sequences of identical symbols.
– Encode them as the symbol followed by its count (e.g.,
"AAA" → "3A").
Advantages:
– Very effective for datasets with long runs of repeated
values.
Applications:
– Image compression (e.g., bitmap images)
– Simple text data compression
 Comparison of Data Reduction
Type of
Techniques
Lossless
Algorithm Best Used For
Compression /Lossy
Time series data, Long-term
Turning Point Approximation Lossy ECG/EEG with significant
changes
Continuous ECG/EEG with low
AZTEC Compression Lossless
variability
Image/Video, High-frequency
Differential
Fan Lossless ECG/EEG with minimal
Encoding
variation
Text/General data, General
Huffman Statistical Encoding Lossless
ECG/EEG signal compression
Specialized data (e.g., fax),
Modified Huffman Statistical Encoding Lossless Complex ECG/EEG with
repeating patterns
Text, Images, Stable ECG/EEG
Run Length Repetition-based Lossless
segments (e.g., sleep, rest)
 Applications of Data Reduction
Techniques
• Data Transmission: Compress data for faster
transmission over networks (e.g., ZIP, JPEG).
• Storage Optimization: Reduce the space
required to store files (e.g., database
compression).
• Signal Processing: Compress sensor data or
audio signals for efficient storage and
transmission.
 Conclusion

• Data reduction techniques are essential for

efficient ECG and EEG signal analysis.
• Different algorithms are suited for different
aspects of signal characteristics.
• Proper compression enables faster processing,
reduced storage needs, and better data
transmission for real-time applications.
• As ECG and EEG signal data grows, efficient data
reduction will be vital for improving patient care
and enabling real-time clinical decision-making.