Chirag, DSML

1. Predic ve Analy cs for IoT

Unit – I

Challenges and opportuni es in predic ve analy cs for IoT data, Characteris cs of IoT Data: Real me streaming,
high frequency, and historical data. Advanced Data Preprocessing: handling missing and noisy data (e.g., imputa on,
filtering). Feature engineering for IoT: Aggrega on, me-lag features, Fourier transforms.

Unit – II

Exploring IoT data with Python: Using NumPy, and matplotlib for IoT data explora on. Cleaning and standardizing
IoT data: Handling missing values, outlier detec on, and normaliza on. Advanced data explora on techniques:
Correla on analysis, data distribu on visualiza on, and anomaly detec on.

Feature Engineering Overview: Importance and techniques in the context of IoT. Feature extrac on from IoT data:
Aggrega ng me-series data, crea ng derived features, and handling categorical data.

Unit – III

Feature selec on: Techniques like Recursive Feature Elimina on (RFE) and correla on-based feature selec on.
Training predic ve models: Supervised learning for fault predic on and maintenance scheduling.

Analyzing model performance: Metrics such as precision, recall, F1-score, and confusion matrix. Scaling predic ve
analy cs solu ons: Deploying models using Azure Machine Learning and monitoring performance.

Unit – IV

IoT Applica ons: Smart Ci es: Traffic flow predic on, energy op miza on. Smart Agriculture: Crop yield predic on,
irriga on forecas ng. Healthcare: Predic ve pa ent monitoring systems.

Security and Privacy in IoT Analy cs: Challenges in securing IoT data pipelines. Blockchain for IoT data sharing.
 2. Data Visualiza on & Interpreta on

Unit – I

Introduc on to Data Visualiza on: Importance and evolu on of data visualiza on in decision-making and analy cs.
Principles of Visualiza on: Cogni ve principles of visual percep on: Pre-a en ve processing, Gestalt principles, and
effec ve use of color.

Visual Encodings: Marks and channels, data types (categorical, ordinal, quan ta ve), choosing visual encodings for
specific data types.

Unit – II

Design Principles: Edward Tu e’s design philosophy, the balance between simplicity and detail, avoiding visual
clu er, and data-ink ra o. Types of Charts: Overview and appropriate usage of bar charts, line charts, sca er plots,
histograms, pie charts, and heatmaps.

Exploratory Data Analysis (EDA): Descrip ve sta s cs (mean, variance, skewness, kurtosis), detec ng outliers.

Unit – III

Advanced Visualiza on Techniques: Mul dimensional data visualiza ons (pair plots, parallel coordinates, and
bubble charts). Hierarchical data visualiza ons (tree maps, dendrograms). Timeseries visualiza ons (line graphs, area
charts, and interac ve melines).

Data Labeling and Annota ons: Best prac ces for annota ons, legends, and tles to make visualiza ons self-
explanatory.

Unit – IV

Overview of Power BI: Introduc on to Power BI, its architecture, and its components (Power Query, Power Pivot, and
Power View). Interface: understanding the Power BI workspace, menus, and naviga on panel, connec ng Power BI to
various data sources like Excel, CSV, and databases. Power Query Editor: cleaning and transforming data (removing
duplicates, handling null values).

Types of Visualiza ons: Bar charts, line charts, pie charts, tables, matrix views, and maps. Customizing Visuals:
forma ng visual elements, adding labels, tool ps, and colors.
 3. Natural Language Processing

Unit-I

Overview of Natural Language Processing: Origins of NLP, Challenges of NLP, Stages of NLP, Applica ons such as
informa on extrac on, ques on answering, and machine transla on. The problem of ambiguity, Why NLP is difficult.
Word Level Analysis: Regular Expressions, Finite-State Automata, Morphology Parsing, Spelling Error Detec on and
Correc on, Words and Word Classes, Part-of-Speech Tagging.

Regular Expressions: Regular Expressions, Automata, Similarity Computa on: Regular Expressions, pa erns, FA,
Formal Language, NFSA, Regular Language and FSAs, Raw Text Extrac on and Tokeniza on, Extrac ng Terms from
Tokens, Vector Space Representa on and Normaliza on, Similarity Computa on in Text.

Unit-II

Morphology and Finite-State Transducers: Inflec on, Deriva onal Morphology, Finite- State Morphological Parsing,
The Lexicon and Morphotac cs, Morphological Parsing with Finite State Transducers, Combining FST Lexicon and
Rules, Lexicon-free FSTs: The Porter Stemmer, Human Morphological Processing.

Matrix Factoriza on and Topic Modeling: Introduc on, Singular Value Decomposi on, Nonnega ve Matrix
Factoriza on, Probabilis c Latent Seman c Analysis, Latent Dirichlet Alloca on.

Unit-III

Computa onal Phonology and Text-to-Speech: Speech Sounds and Phone c Transcrip on, The Phoneme and
Phonological Rules, Phonological Rules and Transducers, Advanced Issues in Computa onal Phonology, Machine
Learning of Phonological Rules, Mapping Text to Phones for TTS, Prosody in TTS .

Probabilis c Models of Pronuncia on and Spelling: Dealing with Spelling Errors, Spelling Error Pa erns, Detec ng
NonWord Errors, Probabilis c Models, Applying the Bayesian method to spelling, Minimum Edit Distance, English
Pronuncia on Varia on, The Bayesian method, Pronuncia on in Humans.

Unit-IV

N-gram Language Models: The role of language models. Simple N-gram models. Es ma ng parameters and
smoothing. Evalua ng language models. Smoothing, Backoff, Deleted Interpola on, N-grams for Spelling and
Pronuncia on, Entropy.

Markov Model and POS Tagging: Overview of Hidden Markov Models, Parameter es ma on, Informa on sources in
tagging: Markov model taggers, The Viterbi Algorithm Revisited,

Word Classes and Part-of-Speech Tagging: Tagsets for English, Part of Speech Tagging, Rule-based Part-ofspeech
Tagging, Stochas c Part-of-speech Tagging, Transforma on- Based Tagging.
 4. Reinforcement Learning

Unit-I

Introduc on: Basics of RL, RL task formula on (ac on space, state space, environment defini on), Defining RL
framework, The Reinforcement Learning problem: evalua ve feedback, nonassocia ve learning, Rewards and
returns, Markov Decision Processes, Value func ons, op mality and approxima on.

Bandit Problems: Explore-exploit dilemma, Binary Bandits, Learning automata, explora on schemes Dynamic
programming: value itera on, policy itera on, asynchronous DP, generalized policy itera on.

Unit-II

Monte-Carlo methods and Temporal Difference Learning: Monte Carlo: Predic on, Es ma on of Ac on values,
Control and Control without Exploring Starts, Policy evalua on, Roll outs, On Policy and Off Policy learning, Temporal
Difference Predic on: TD (0), Op mality of TD (0), SARSA: On Policy TD control, Q-learning: Off Policy TD control, R-
learning, Games and a er states.

Eligibility traces: n-step TD predic on, TD (lambda), forward and backward views, Q(lambda), SARSA (lambda),
replacing traces and accumula ng traces.

Unit-III

Func on Approxima on: Value predic on, gradient descent methods, linear func on approxima on, Control
algorithms, Fi ed Itera ve Methods Policy Gradient methods: non-associa ve learning - REINFORCE algorithm, exact
gradient methods, es ma ng gradients, approximate policy gradient algorithms, actor-cri c methods.

Deep Reinforcement Learning: Deep Q-Networks, Double Deep-Q Networks (DQN, DDQN, Dueling DQN, Priori zed
Experience Replay)

Unit-IV

Hierarchical RL: MAXQ framework, Op ons framework, HAM framework, Inverse reinforcement learning, Maximum
Entropy Deep Inverse Reinforcement Learning, Genera ve Adversarial Imita on Learning, Recent Trends in RL
Architectures, Applica ons of Reinforcement learning: NLP, healthcare, finance, educa on, robo cs, games computer
vision.
 5. Informa on Security and Privacy

Unit-I

Founda ons of Informa on Security and Privacy: Overview of Informa on Security - History, defini ons, and
importance in today’s digital world.

Security Models & Principles: Confiden ality, integrity, availability (CIA Triad); defence in depth; risk management
(NIST framework).

Threat Modelling & A ack Vectors: Social engineering, malware, insider threats, APTs. Fundamentals of Privacy:
Defini ons, privacy principles, and the interplay between security and privacy.

Unit-II

Cryptography, Network Security, and Access Control: Cryptographic Primi ves and Protocols -
Symmetric/asymmetric encryp on, hash func ons, digital signatures, PKI.

Secure Communica on Protocols: TLS/SSL, IPSec, secure email, VPNs.

Network Security Techniques: Firewalls, intrusion detec on and preven on, segmenta on, zero-trust architectures.
Access Control Models: Discre onary, mandatory, and role-based access control (RBAC).

Unit-III

Privacy, Data Protec on, and Legal/Regulatory Frameworks: Privacy Theories and Models - Contextual integrity,
privacy by design.

Data Protec on Techniques: Data anonymiza on, pseudonymiza on, differen al privacy. Privacy Laws (e.g. GDPR,
HIPAA, CCPA, IT Act 2000, IT Rules 2011, PDP Bill 2019, etc.) and other privacy frameworks.

Ethical Issues in Privacy: Balancing individual privacy rights and organiza onal interests.

Unit-IV

Advanced Topics and Emerging Trends in Informa on Security: Security Management & Incident Response - Security
policy development, audi ng, and compliance frameworks.

Digital Forensics and Cyber Incident Handling: Evidence collec on, analysis techniques, legal considera ons.
Emerging Technologies and Threats: Cloud security, IoT, AI-based threats, blockchain security.

Future Direc ons: Research trends, advanced cryptographic protocols (post-quantum cryptography), privacy
enhancing technologies.