Introduction to Machine

Learning in Biological Systems

(ES60011)
 Introduction to Machine Learning
An Overview

• What is Machine Learning?

• Need of Machine Learning
• Types of Machine Learning
• Brief Introduction on Machine Learning Algorithms
• Basic Idea on Machine Learning Steps
• Applications
• Brief Introduction on Biological Systems
• Machine Learning in Biological Systems
 Need of Machine Learning
Machine learning is applied in various
scenarios, such as:
When human knowledge is
unavailable
e.g.: exploring distant planets like
Mars
When humans are unable to Emotion Pattern
Evaluation
articulate their skills
Personalised
e.g.: recognizing speech patterns Healthcare
When solutions need to be tailored
e.g.: designing individualized
treatment plans in healthcare
When decisions rely on vast
datasets
e.g.: analyzing genomic information Space
Navigation

Big Data
Handling
 What are ML techniques

Machine leaning (ML) techniques enables systems to learn from

experience (read data). ML refers to systems ability to acquire and
integrate knowledge through large-scale observations and to improve
and extend by itself learning new knowledge rather than by being
programed with that knowledge (Shapiro 1992)
Types of Machine Learning
 Types of Machine Learning

Machine Learning

Supervised ML Unsupervised ML Reinforcement Learning

Polynomial regression Partial least squares Q-Learning
Random forest (RF) Fuzzy means State-Action-Reward-
Linear regression Singular value State-Action (SARSA)
Logistic regression decomposition Deep Q Network (DQN)
Decision trees K-means clustering Deep Deterministic
K-nearest neighbours Apriori Policy Gradient (DDPG)
Naive Bayes Hierarchical
clustering
Principal component
analysis
 Machine Learning: Introduction
Artificial intelligence (AI):
It is computer software that mimics
human cognitive abilities in order to
perform complex tasks that historically
could only be done by humans, such as
decision making, data analysis, and
language translation.

Machine learning (ML):

Machine learning is a subset of AI in
which algorithms are trained on data sets
to become machine learning models
capable of performing specific tasks.

Deep learning:
Deep learning is a subset of ML, in which
artificial neural networks (ANNs) that
mimic the human brain are used to
perform more complex reasoning tasks
without human intervention.
 Machine Learning Algorithms
Supervised (inductive) learning

– Given: training data + desired outputs (labels)

Supervised
Learning

Regression Classification

•• Given (x1, y1), (x2, y2), ..., (xn, yn)

Learn a function f(x) to predict y given x
•• Given (x1, y1), (x2, y2), ..., (xn, yn)
Learn a function f(x) to predict y given x – y is categorical ==
– y is real-valued == regression classification

•x can be multi-dimensional
– Each dimension corresponds to an attribute
Evaluation metrics - For Classification Problem
•Analysis:
•Precision
•Recall
•Accuracy
•F1-score
•Etc…
True Positive (TP)
The predicted value matches the actual value.
The actual value was positive and the model
predicted a positive value.
True Negative (TN)
The predicted value matches the actual value.
The actual value was negative and the model
predicted a negative value.
False Positive (FP) – Type 1 error
The predicted value was falsely predicted. The
actual value was negative but the model
predicted a positive value.
False Negative (FN) – Type 2 error
The predicted value was falsely predicted. The
actual value was positive but the model
predicted a negative value.
 Evaluation metrics

Accuracy= (TP+TN)/(TP+TN+FP+FN)

Precision P = TP/(TP+FP)

Recall/Sensitivity R= TP/(TP+FN)

Specificity S = TN/(TN+FP)

F-measure F=2*(P*R)/(P+R)
[harmonic mean of precision and recall]
 Evaluation metrics

Accuracy= (TP+TN)/(TP+TN+FP+FN)
= 77.51 %
 Numericals
1. Suppose a regression model follows y=mx+c, if the dataset (X,y) with
X=[1,2,3,4,5] and y = [2,4,6,8,10], find the parameters m (slope) and c
(intercept) for this model.
2. If a binary classification model has 90 true positives, 30 false positives,
20 false negatives, and 160 true negatives, calculate the accuracy,
precision, recall, and F1-score.
 Machine Learning Algorithms
Unsupervised learning
– Given: training data (without desired outputs)

Unsupervised
Learning

Clustering Dimensionality
Reduction
Given x1, x2, ..., xn (without labels)
Output hidden structure behind the x’s Given x1, x2, ..., xn (without labels)
Reduces the number of variables to get
the exact information
 Numericals
Given the following points: (1, 1), (1, 4), (4, 1), and (4, 4), if we want to
cluster them into 2 clusters, what are the initial centroids and the final
centroids after one iteration of K-means clustering?
 Machine Learning Algorithms
Reinforcement Learning
Discovers data through a process of trial and error and then decides what
action results in higher rewards.
Major components: the agent/learner/decision-maker, the environment, and
the actions.
Reinforcement Learning

Given a sequence of states and actions with (delayed) rewards, output a policy
Policy is a mapping from states actions that tells you what to do in a given state
 Machine Learning Steps
• Data Collection
• Data Cleaning and
Preprocessing
• Exploratory Data
Analysis (EDA)
• Feature Engineering
• Model Selection
• Model Training
• Model Evaluation
• Hyperparameter Tuning
• Model Deployment
• Monitoring and
Maintenance
• Documentation and
Reporting
 Applications
Supervised ML Unsupervised ML Reinforcement Learning

• Image Classification • Customer • Robotics

• Spam Detection Segmentation • Game Playing (e.g.,
• Sentiment Analysis • Anomaly Detection Chess, Go, and
• Predictive • Market Basket Video Games)
Maintenance Analysis • Autonomous
• Medical Diagnosis • Document Clustering Vehicles
• Fraud Detection • Dimensionality • Industrial Automation
• Speech Recognition Reduction • Personalized
• Recommendation • Image Compression Recommendations
Systems • Gene Sequence • Financial Trading
• Customer Churn Analysis • Healthcare
Prediction • Topic Modeling Treatment Planning
• Stock Price • Social Network • Natural Language
Prediction Analysis Processing
• Handwriting • Recommender • Smart Grid
Recognition Systems Management
• Credit Scoring • Data Visualization • Supply Chain
• Feature Learning Optimization
• Advertising Bidding
• Dynamic Pricing
 Machine Learning in Biological Systems
Ecology and Environmental Biology

Species Identification: ML models are used to identify species from images, sounds,
and other data, aiding in biodiversity studies.

Ecosystem Modeling: ML helps in modeling ecological systems and predicting the

impact of environmental changes on ecosystems.
Machine Learning in Biological Systems

Ref.: Auslander, N., Gussow,

A. B., & Koonin, E. V. (2021).
Incorporating machine
learning into established
bioinformatics frameworks.
International journal of
molecular sciences, 22(6),
2903.
Machine Learning in Biological Systems

Ref.: Auslander, N., Gussow, A. B., & Koonin, E. V. (2021). Incorporating machine learning into established bioinformatics frameworks. International
journal of molecular sciences, 22(6), 2903.
 Machine Learning in Biological Systems
• Machine learning is increasingly being utilized to
understand and model biological systems due to its ability
to handle and analyze large, complex datasets.
• Genomics:
• Identifying patterns in DNA sequences,
• Predicting gene expression,
• Understanding genetic variations associated with diseases.
• Proteomics:
• Analyzing protein structures and functions,
• Predicting protein interactions,
• Identifying biomarkers for diseases.
• Drug Discovery:
• Predicting the efficacy and toxicity of new compounds,
• Identifying potential drug targets,
• Optimizing drug design.
 Machine Learning in Biological Systems
• Medical Imaging:
• Enhancing the analysis of medical images (e.g., MRI, CT scans) for disease
diagnosis and treatment planning.
• Systems Biology:
• Modeling complex biological networks to understand cellular processes, metabolic
pathways, and disease mechanisms.
• Personalized Medicine:
• Tailoring medical treatments to individual patients based on their genetic and
phenotypic information.
• Epidemiology:
• Predicting the spread of infectious diseases,
• understanding risk factors,
• planning public health interventions.
• Neuroscience:
• Analyzing neural activity data to understand brain function, cognitive processes,
and neurological disorders.
• Machine learning models can uncover hidden patterns and relationships within
biological data, leading to new insights and advancements in biological research
and healthcare.
 Challenges and Future Directions
•(e.g.,
Data Integration: Combining data from different sources and types
genomic, proteomic, clinical) remains a challenge.

•crucial
Interpretability: Understanding how ML models make decisions is
for their acceptance in critical fields like medicine.

•scalable
Scalability: Handling large-scale biological data efficiently requires
ML methods.

•implications
Ethical Considerations: Issues such as data privacy and the ethical
of ML-driven decisions need careful consideration.

• Overfitting
• Underfitting