A
abstract data types (ADT)
exploring 43
tree 50
abstract data types (ADT), stacks
practical example 47
time complexity 47
abstract data types (ADT), vector
time complexity 44
actionable rules
examples 176
activation functions 262
hyperbolic tangent (tanh) function 268, 269
ReLU activation function 266
step function 264
adjacency list
constructing 126
Advanced Encryption Standard (AES) 424
advanced lossless compression formats 404
GZIP compression 405
LZO compression 404
snappy compression 404
advanced sequential modeling techniques
AI spring 252
AI winter 252
approximate algorithm 23
coding phase 6
compute-intensive algorithms 10
data-intensive algorithms 10
design phase 5
deterministic algorithm 22
development environment 7
exact algorithm 23
explainability 23
performance analysis 12
performance, estimating 15
randomized algorithm 22
selecting 21
validating 22
algorithm design techniques
compute dimension 12
algorithmic ethics 467
bias and discrimination 467
privacy 467
problems 468
solution factors 469
algorithmic ethics, solution factors
inconclusive evidence, considering 470
misguided evidence 470
traceability 470
unfair outcomes 471
algorithmic solutions
challenges 458
unexpected disruption 458, 459
algorithmic strategies 87
divide-and-conquer strategy 87
dynamic programming strategy 90
greedy algorithms 92
Amazon Web Services (AWS) 7, 218, 439
Amdahl’s law 444
Apache Spark
divide-and-conquer strategy, applying 88-90
large-scale algorithms, processing with 454, 455
reference link 88
using, in cloud computing 452
Apache Spark architecture
cluster manager 453
driver program 452
executors 453
worker node 453
append()
elements, adding with 29
elements, removing with 30
apriori algorithm 178
limitation 179
AP systems 397
Arithmetic Logic Units (ALUs) 448
association analysis algorithms 178
apriori algorithm 178
apriori algorithm, limitation 179
FP-growth algorithm 178
FP-growth algorithm, using 183-185
frequent pattern growth (FP-growth) algorithm 179
frequent patterns, mining 182, 183
Association for Computing Machinery (ACM) 24
association rules mining 172, 175
actionable rules 176
association analysis algorithms 178
inexplicable rules 176
lift 178
ranking rules 176
support measure 177
trivial rules 175
types of rules 175
asymmetric encryption 424
public key infrastructure (PKI) 427, 428
SSL/TLS handshaking algorithm 425-427
challenges 357
in neural network 353
key aspects 355
overview 356
autoencoder 346
coding 348
environment, setting up 349
reconstruction phase 348
training phase 348
autoencoder, environment
compilation 349
data preparation 349
model architecture 349
prediction 350
training phase 350
average pooling 280
B
backend engines, Keras 271
backpropagation through time (BPTT) 325, 326
bag-of-words-based (BoW-based) 289
betweenness centrality 124
bias model
Bidirectional Encoder Representations from Transformers (BERT) 366
big data 86
functions 31
binary classifier 201
performance analysis 73
binary tree 51
black box algorithm 460
versus white box algorithm 460
black swan event
challenges and opportunities 473
defining 472
practical application 473
characteristics 428
breadth-first search (BFS) 116, 126
adjacency list, constructing 126
breadth-first search (BFS), algorithm implementation 127
initialization 127
loop 127
specific searches, using 129, 130
brute-force strategy
bubble sort algorithm 57
performance analysis 60
bubble sort algorithm, time complexity
best case 60
worst case 60
C
candidate cell state 334
candidate-generation phase 179
CAP theorem 394
AP systems 397
availability 395
CA systems 397
connecting 395
consistency 395
CP systems 398
partition tolerance 395
case conversion 294
CA systems 397
centrality measures 119
betweenness 120
closeness 121
eigenvector centrality 121
fairness 121
centrality metrics
calculating, with Python 122
graph, crafting 122
graph, visualizing 123
libraries and data, setting 122
Central Processing Units (CPUs) 439
Certification Authority (CA) 427
cipher 411
designing 414
suite 411
text 411
classification algorithms 193, 233, 234
feature engineering, with data processing pipeline 236, 237
historical dataset, exploring 235, 236
regression algorithms 234
regressors challenge, presenting 235
regressors challenge, problem statement 235
Classification And Regression Tree (CART) algorithm 93
classifiers 193
versus regressors 193
classifiers challenge 194
evaluation 200
feature engineering, with data processing pipeline 195
feature normalization 199
logistic regression, using 226, 227
Naive Bayes theorem, using 233
Random Forest algorithm, using 222, 223
SVM algorithm, using 229
classifiers phase
cleaning data 294
case conversion 294
numbers, handling 295
punctuation removal 294
stopword removal 296
white space removal 296
closeness 121
closeness centrality 124
Cloud and Algorithmic Scalability
elasticity 87
cloud computing
Apache Spark, using 452
large-scale algorithms, using 455
Cloud for distributed model training
advantages 455
clustering algorithms 149
Euclidean distance 151
Manhattan distance 152
quantify similarities 150
clusters
application 166
creating, with DBSCAN in Python 164, 165
evaluating 166
clusters, government use cases
crime-hotspot analysis 166
demographic social analysis 166
clusters, market research
customer categorization 167
market segmentation 166
targeted advertisements 167
collaborative filtering engines 374, 376
issues 377
compilation 349
complexity theory 12
components, attention mechanism
contextual relevance 355
prioritized focus 355
symbol efficiency 355
compression algorithms
using 406
computational ethics 467
compute-intensive algorithms 10
Compute Unified Device Architecture (CUDA)
Bottom of Form 448
GPU architectures, in parallel computing 447
parallel processing, in LLMs 449, 450
precision 202
recall 202
constant time (O(1)) complexity 18, 19
content-based recommendation engines 374, 375
unstructured documents, determining 375, 376
contextual relevance 355
context vector (c2) 357
Converging Iterations 13
convolution neural networks (CNNs) 279
convolution 280
pooling 280
cost function 225
CP systems 398
crime-hotspot analysis 166
Cross-Industry Standard Process for Data Mining (CRISP-DM) 144, 211, 430, 471
reference link 144
cryptanalysis 411
cryptographic hash function
application 422
implementing 419
implementing, with MD5 420, 421
implementing, with SHA 421, 422
cryptographic techniques
asymmetric encryption 424
symmetric encryption 423
types 417
CUDA Deep Neural Network (cuDNN) library 271
Cyclic Redundancy Check (CRC) 413
D
data algorithms 394
CAP theorem 394
CAP theorem, connecting 395
data compression, connecting 395
distributed environment storage 394, 395
data cleaning
data compression
connecting 395
data compression algorithm
decoding 398
lossless compression techniques 398
DataFrame 39
subset, creating 39
using 38
data-intensive algorithms 10
data management in AWS 405
benefits, quantifying 406, 407
CAP theorem, applying 405, 406
compression algorithms, using 406
data preparation 349
data processing pipeline 195
data representation
for sequential models 317
decision tree classification algorithm 212-215
strengths and weaknesses 216
use cases 216
decoder 352
decryption 411
deep learning technology stack 271
deep model
using, to create LLMs 367
deep neural network 256
degree centrality 123
degree of suspicion (DOS) 137-139
Deletion Operations 42
demographic social analysis 166
dendrogram 161
Density-based spatial clustering of applications with noise (DBSCAN) 163
used, for creating clusters in Python 164, 165
depth-first search (DFS) 126, 130-132
designed algorithm
concerns 80
correctness 81
scalability 86
deterministic algorithm 22
need for 37
dictionary-based compression LZ77 403
example 403
versus Huffman 403
digital certificate 427
dimensionality reduction 167
feature aggregation 167
feature selection 167
directed graph (DiGraph) 113, 277
distributed computing 454
distributed environment storage 394
Distributed Ledger Technology (DLT) 428
Distributed Shared Memory (DSM) 442
Diverging Iterations 13
divide-and-conquer strategy 87
applying, to Apache Spark 88-90
domain validation 427
downsampling 280
performing 280
driver machine 452
dynamic programming strategy 90
characteristics 91
components 91
conditions 91
E
ego-centered network 114
applications 115
basics 114
one-hop neighbors 114
two-hop neighbors 114
Eigen 271
eigenvector centrality 121, 124
Elastic Compute Cloud (EC2) 439
Elastic Load Balancing (ELB) 439-442
encryption 411
End-Of-Sentence (<EOS>) 352
ensemble boosting
versus Random Forest algorithm 221
ensemble methods 217
gradient boosting, implementing with XGBoost algorithm 218, 220
Euclidean distance 151
event types
dependent 230
independent 230
mutually exclusive 230
exact algorithm 23
explainability
of algorithm 23
F
fairness 121
False Positive Rate (FPR) 207
fault tolerance 440
feature engineering 195
data, importing 196
dataset, into testing portion 199
dataset, into training portion 199
features and label, specifying 199
with data processing pipeline 236, 237
feature normalization 199
Feature Vector 1 284
Feature Vector 2 284
feedforward neural networks (FFNNs) 362
filter 280
filter phase 179
First In, First Out (FIFO) principle 130
First-In, Last-Out (FILO) principle 45
fixed-length code 402
Flat Iterations 13
forget gate 333
fraud detection with deep learning, case study 283
fraud detection with SNA 133
simple fraud analytics, conducting 135, 136
watchtower fraud analytics methodology, presenting 136
fraud (F) 133
frequent pattern growth (FP-growth) algorithm 179
full tree 52
Functional API 272
functional model
selecting 276
functional requirements 6
G
Garbage-in, Garbage-out (GIGO) 470
gating mechanism 330
Generative Adversarial Networks (GANs) 281
Generative Pre-trained Transformer (GPT) 366
global explainability strategy 461
Google Cloud Platform (GCP) 7
gradient boosting
implementing, with XGBoost algorithm 218, 220
gradient boost regression algorithm 243
using, for regressors challenge 244
gradient descent
graphs
ego-centered network 114
graph theory 112
graph traversals 125
depth-first search (DFS) 130-132
greedy algorithms 92
characteristics 92
hidden cell, updating 331, 332
running, for multiple timesteps 332
update gate, implementing 331
GZIP compression 405
H
hidden cell
hierarchical clustering 161, 162
algorithm, coding 162
historical dataset
homophily principle 133
Huffman
versus dictionary-based compression LZ77 403
example 399
implementing, in Python 399-402
Huffman tree 399
human brain
axon 251
dendrites 251
synapse 251
hybrid recommendation engines 374, 378
recommendations, generating 380
recommendation system, evolving 381
reference vectors, generating 379
similarity matrix, generating 379
hybrid recommendation systems 382
double-edged sword of social influence 382, 383
hyperbolic tangent (tanh) function 268, 269
hyperparameters
defining 272
I
information bottleneck 353
in-memory processing 454
inner loop 60
Insertion Operations 41
insertion sort algorithm 61
performance analysis 62
intercluster distance 150
performance analysis 74
Interquartile Range (IQR) method 156
intracluster distance 150
Inverse Document Frequency (IDF) 300
itemset 174
K
Keras 270
architecture 271
backend engines 270
hyperparameters 272
low-level layers, of deep learning stack 271
Microsoft Cognitive Toolkit (CNTK) 271
reference link 270
TensorFlow 271
Theona 271
kernel 280
k-means clustering algorithm 155
hierarchical clustering 161
initialization 155
limitation 160
logic 155
stop condition 157
L
labeled data 193
Language models (LMs) 364
Large Language Models (LLMs) 345, 346, 364, 365
deep model, using 367
wide model, using 367
large-scale algorithms 438
characterizing 439
fault tolerance 440
parallelism 440
performant infrastructure, characterizing 439
processing, with Apache Spark 454, 455
using, in Cloud computing 455
Last In, First Out (LIFO) principle 45, 130
layered deep learning architectures 255
intuition, developing for hidden layers 256, 257
mathematical foundation of neural networks 258, 259
optimal number of hidden layers 257
Leaky ReLU activation function 267
default value 267
parametric ReLU 268
randomized ReLU 268
lift 178
linear discriminant analysis (LDA) 167
linear programming 103
capacity, planning with 104-107
constraints, specifying 104
objective function, defining 104
problem, formulating 104
linear regression 237
gradient boost regression algorithm 243
gradient boost regression algorithm, using for regressors challenge 244
regression tree algorithm 242
regressors, evaluating 239, 240
simple linear regression 238
usage 242
using, for regressors challenge 241
weaknesses 242
performance analysis 72
linear time (O(n)) complexity 19
lists 26
modifying 29
time complexity 31
using 27
load balancing
combining, with elasticity 441, 442
fault tolerance 440
local explainability strategy 461
Local Interpretable Model-Agnostic
logarithmic time (O(logn)) complexity 20, 21
logistic function 225
logistic regression 223
assumption 224
cost function 225
for classifiers challenge 226, 227
loss function 225
need for 226
relationship, establishing 224
Long Short-Term Memory (LSTM) 332, 333, 364
candidate cell state 334
cell state 333
forget gate 333
hidden state 333
memory state, calculating 335
sequential models, coding 337
working, with multiple timesteps 337
loss function 225
lossless compression techniques 398
advanced lossless compression formats 404
LSTM, sequential models
data, preparing 339
incorrect predictions, viewing 343
model, training 342
LZO compression 404
M
machine learning algorithm 460
machine learning explainability 461
strategies, presenting 461, 462
Machine Learning (ML) 148, 252
machine learning model, security concerns 430
masquerading, avoidance 432
Manhattan distance 152
Man-in-the-Middle (MITM) attacks 430, 431
many-to-many sequence models 316, 317
many-to-one sequence models 315, 316
market basket analysis 174
market segmentation 166
Matplotlib 9
reference link 9
matrices 43
Matrix factorization methods 382
matrix operations 42
max pooling 281
and SHA, selecting between 422
using 421
Mean Absolute Error (MAE) 390
Mean Squared Error (MSE) 242
Measure Of Similarity (MOS) 284
memorization 369
memory state
calculating 335
merge sort algorithm 63
pseudocode overview 64
Python implementation 64
merging phase 63
Microsoft Cognitive Toolkit (CNTK) 271
reference link 271
Miles per Gallon (MPG) 235
model architecture 349
Modified National Institute of Standards and Technology (MNIST) 348
reference link 348
multilayer neural network 255
hidden layer 256
input layer 255
output layer 256
multilayer perceptron 255
multi-resource processing
N
Naive Bayes algorithm 230
Naive Bayes' theorem 230
addition rules for OR events 232
for classifiers challenge 233
general multiplication rule 232
multiplication rules for AND events 231
probabilities, calculating 231
National Research Council (NRC) 410
natural language processing (NLP) 55, 289, 290, 312, 346, 455
applications 309
exploring 365
negative outcomes
scoring 136
network analysis 112
network analysis theory 115
centrality measures 119
centrality metrics, calculating with Python 122
shortest path 116
Social Network Analysis (SNA) 125
networkx Python package
reference link 113
activation function 260
anatomy 259
attention mechanism 353
basic idea 353
convolution neural networks (CNNs) 279
cost function 259
example 354
fraud detection case study 283
Generative Adversarial Network (GANs) 281
input data 260
layered deep learning architectures 255
layers 259
loss function 259
optimizer 259
output 363
perceptrons 252
tools and frameworks 270
training 259
transformer architecture 362
types 279
weights 260
NLP terminology 290
corpus 290
Named Entity Recognition (NER) 291
normalization 291
stemming and lemmatization 291
stop words 291
tokenization 291
Non-Deterministic Polynomial (NP) 83
versus NP-complete 85
versus NP-hard 85
versus polynomial (P) 85
non-fraud (NF) 133
non-functional requirements 6
NP-complete 84
NP-hard 85
versus polynomial (P) 85
NumPy 9
reference link 9
O
one-hop neighbors 114
one-to-many sequence models 313-315
characteristics 314
ordered tree 52
outer loop 60
P
PageRank algorithm
presenting 100
problem, defining 100
pandas 9
reference link 9
Series-based data structures 38
parallel computing
GPU architectures 447
limitations 444
Parallelism 440
partitions 452
passes 60
perceptrons 252
perfect tree 52
performance estimation, algorithm
average case 16
best case 15
worst case 16
performant infrastructure, characterizing for large-scale algorithm
elasticity 439
Personally Identifiable Information (PII) 411
Pip Installs Python 7
plain text 411
polynomial algorithm 82
polynominal (P) 83
pooling 280
advantages 280
average pooling 281
max pooling 281
practical application area 383
Amazon recommendation system 384
data-driven recommendation 383
principal component analysis (PCA) 167-171
association rules mining 172
examples 173
limitations 172
market basket analysis 173, 174
Public Key Infrastructure (PKI) 427, 428
punctuation removal 294
PyPI 7
Python 7
Huffman coding, implementing 399-402
reference link 7
used, for calculating centrality metrics 122
variables, swapping 56
Python built-in data types
DataFrames, using 38
exploring 26
lists 26
lists, using 27
matrices 42
series, using 38
time complexity, of dictionaries 35
Python built-in data types, lists
elements, adding with append() 29
elements, adding with pop() 30
elements, removing with pop() 30
iteration 29
list indexing 27
modifying 29
negative indexing 28
nesting 28
range function 30
time complexity 31
Python built-in data types, matrices
Big O notation 43
matrices 43
matrix operations 42
Python Notebook
using 9
Q
quadratic time (O(n2)) complexity 19, 20
time complexity 49
usage 49
R
Random Forest algorithm 220, 221
using, for classifiers challenge 222, 223
versus ensemble boosting 221
randomized algorithm 22
range function 30
ranking rules 176
Receiving Operating Curve (ROC) 207
recommendation engine
correlation 389
creating 385
data, merging 386
descriptive analysis 387
framework, setting up 385
model, evaluating 390
movies, correlating 389
structuring 387
user feedback, retraining 390
recommendation system 374
cold start problem 381
collaborative filtering engines 376
content-based recommendation engines 375
data sparsity problem 382
hybrid recommendation engines 378
limitations 381
metadata requisites 382
types 374
Recurrent Neural Network (RNN) 318, 352
architecture 318
Backpropagation through time (BPTT) 325, 326
output, calculating for each timestep 324
training, at first timestep 320
training, for whole sequence 322, 323
recursive function 91
regression tree algorithm 242
using, for regressors challenge 243
regressors 193
versus classifiers 193
regressors challenge
gradient boost regression algorithm, using 244
linear regression, using 241
presenting 235
problem statement 235
regression tree algorithm, using 243
regular RNNs 360
ReLU activation function 266
Leaky ReLU 267
Resilient Distributed Datasets (RDDs) 452
restaurant review sentiment analysis, case study 306
results, analyzing 308
text data, converting into numerical features 307
text data, preprocessing 307
RNNs, architecture
input variable, characteristics 319, 320
memory cell and hidden state 318, 319
RNNs, for whole sequence
weight parameter matrix, combining 323, 324
Root Mean Square Error (RMSE) 240, 390
Rotation 13 (ROT13) 416
S
scaling 199
scikit-learn 9
reference link 9
SciPy ecosystem 8
SciPy ecosystem, packages
Matplotlib 9
NumPy 9
pandas 9
scikit-learn 9
searching algorithm 70
binary search 72
Secure Hashing Algorithm (SHA) 421, 422
Secure Sockets Layer (SSL)/Transport Layer Security (TLS) 425
security requirements 411
data sensitivity 413
entities, identifying 412
security goals, establishing 412
Selection Operations 41
selection sort algorithm 68, 69
performance analysis 69
self-attention 357
attention weights 358
bidirectional RNNs 359
regular RNNs 360
thought vector 360
training, versus inference 360, 361
sensitive data 413
Sequence-to-Sequence (Seq2Seq) models 316, 345, 346, 351
encoder 352
information bottleneck 353
thought vector 352
tokens 352
writer 352
Sequential API 272
types 312
sequential data, types
Spatial-Temporal Data 312
Textual Data 312
Time Series Data 312
sequential model
coding 337
data representation 317
selecting 276
types 313
sequential model, types
many-to-many sequential models 316, 317
many-to-one sequential models 315, 316
one-to-many sequence models 313-315
series
using 38
need for 37
time complexity analysis 37
performance analysis 68
shortest path 116
density 118
neighborhood, creating 116
triangles 117
Siamese neural networks 283
sigmoid function 225, 264, 265
similarity scores
interpreting 305
simple fraud analytics
simple graph 113
simple linear regression 238
simple neural network 256
snappy compression 404
Social Network Analysis (SNA) 125
softmax function 269
sort algorithm
selecting 70
sorting algorithms 56
bubble sort algorithm 57
insertion sort algorithm 61
merge sort algorithm 63
variables, swapping in Python 56
space complexity analysis 13
Spatial-Temporal Data 312
Special Interest Group on Knowledge Discovery (SIGKDD) 24
Speedup 445
splitting phase 63
practical example 47
time complexity 47
usage 49
step function 264
stopword removal 296
subset, of DataFrame
column selection 40
time complexity analysis, for sets 41, 42
substitution-based ciphers
cryptanalysis 416
presenting 414
ROT13 416
supercomputers 442
supervised machine learning 188, 189
classifiers, versus regressors 193
conditions, enabling 192
support measure 177
Support Vector Machines (SVMs) 187
support vectors 228
for classifiers challenge 229
Naive Bayes algorithm 230
symbol efficiency 355
symmetric encryption
advantages 424
issues 424
using 423
T
targeted advertisements 167
Tay Twitter AI bot
failure 459
3D tensor 277
basic concepts, presenting 277, 278
matrix 277
rank 277
scalar 277
tensors 277
URL 271
vector 277
tensors 271
Term Document Matrix (TDX) 299
Term Frequency-Inverse Document Frequency (TF-IDF) 300
summary 302
Term Frequency (TF) 300
test document 284
Testing with Concept Activation Vectors (TCAV) 462
text preprocessing in NLP 292
cleaning data 294
text preprocessing techniques, of NLP
language modeling 291
machine translation 291
sentiment analysis 291
word embeddings 291
Textual Data 312
Theona 271
reference link 271
time complexity analysis 13-15
Time Series Data 312
TLS handshake 425
tokens 438
top-secret data 413
transfer learning 281
examples 282
using 281
transformer architecture 346, 361
Transmission Control Protocol/Internet Protocol (TCP/IP) 413
transnational data 174
transposition 417
transposition ciphers 417
Traveling Salesperson Problem (TSP) 79
brute-force strategy, using 94-97
greedy algorithm, using 98, 99
strategies, comparing 99
tree 50
practical examples 52
true document 284
True Positive Rate (TPR) 207
two-hop neighbors 114
U
undirected graph 113
Universal Language Model Fine-Tuning (ULMFiT) 365
unlabeled data 193
unsupervised learning 143, 144
for marketing segmentation 149
practical examples 149
unsupervised learning, in data-mining lifecycle 144, 145
business 145
data 146
data preparation 146
evaluation 147
implementing 331
V
variable-length code 402
variables
swapping, in Python 56
variety 10
vector
time complexity 44
velocity 10
Virtual Private Cloud (VPC) 412
volume 10
W
watchtower fraud analytics methodology
degree of suspicion (DOS) 137-139
negative outcomes, scoring 136
presenting 136
weakest link
weather prediction
weighted graph 113
white box algorithm 460
versus black box algorithm 460
white space removal 296
wide model
using, to create LLMs 367
Word2Vec
advantages 305
disadvantages 306
used, for implementing word embedding 303, 304
Word2Vec() functions
min_count 304
sentences 304
size 304
window 304
workers 304
implementing, with Word2Vec 303, 304
similarity scores, interpreting 305
writer 352
X
XGBoost algorithm
used, for implementing gradient boosting 218
Thanks for purchasing this book!
Do you like to read on the go but are unable to carry your print books everywhere?Is your eBook purchase not compatible with the device of your choice?
Don’t worry, now with every Packt book you get a DRM-free PDF version of that book at no cost.
Read anywhere, any place, on any device. Search, copy, and paste code from your favorite technical books directly into your application.
The perks don’t stop there, you can get exclusive access to discounts, newsletters, and great free content in your inbox daily
Follow these simple steps to get the benefits: