[SYMBOL][A][B][C][D][E][F][G][H][I][J][K][L][M][N][O][P][Q][R][S][T][U][V][W][X][Z]
2-D array
2-D combinations
2-D linear model
4-D Gaussian clustering
10-fold cross-validation
A/B testing
code
suitability of
theory
vs. Bayesian bandits
accuracy, 2nd
activating neural networks
activation function, 2nd, 3rd
activation profiles
activity trends
ad creative optimization
adversarial bandits
advertising.
See click prediction for online advertising.
adview type
AI (artificial intelligence)
algorithms.
See intelligent algorithms.
all value
alpha term
AND function
Apache Hadoop.
See Hadoop framework.
Apache Kafka.
See Kafka log-processing platform.
artificial intelligence.
See AI.
attributes
AUC (area under a curve)
automated method
awk tool
backpropagation
balanced data
Bandit objects
Bayes theorem, 2nd
Bayesian bandits
A/B vs.
factors impacting
Bayesian networks
Bengio, Yoshua
Bernoulli Restricted Boltzmann Machine.
See BRBMs.
beta distribution, 2nd
bias term, 2nd
bid win (or loss) notification
binary classification
binomial distribution
bipartite graph
bootstrapping, 2nd
Bottou
BRBMs (Bernoulli Restricted Boltzmann Machines)
build method
call number
campaigns
capturing data on web
data collection
managing at scale
naïve approach
Kafka log-processing platform
consumer groups
design patterns
evaluating
plus Hadoop
plus Storm
replication in
showing ads online
categorical features
causation, difference from correlation
CF (collaborative filtering)
choices
A/B testing
code
suitability of
theory
vs. Bayesian bandits
Bayesian bandits
A/B vs.
factors impacting
multi-armed bandits
epsilon decreasing strategy
epsilon first strategy
epsilon greedy strategy
extensions to
Cinematch algorithm
classes of intelligent algorithms
artificial intelligence
machine learning
predictive analytics
classification
classifiers
lifecycle of
statistical classification algorithms
structural classification algorithms
credibility of results
fraud detection with logistic regression
from linear to logistic regression
implementing fraud detection
linear regression primer
need for
structural algorithms
with very large datasets
cleverer algorithm
click prediction for online advertising
bidders
decisioning engine
building, complexities of
contextual information
data preparation
decisioning engine model
feature engineering
information about placement
information about user
mapping predicted click-through rate to bid price
model training
exchange and
ad monitoring
ad placement
bid win (or loss) notification
cookie matching
future of real-time prediction
history and background of
with Vowpal Wabbit machine-learning library
model calibration
preparing dataset
testing model
VW data format
click type
click-streams topic, 2nd, 3rd
click-through rate.
See CTR.
cluster centroids, 2nd
clustering, 2nd
codomain
collaborative filtering.
See CF.
colon character
cols value
concepts
confidence interval
confusion matrix
consumer group, 2nd, 3rd, 4th
contextual bandits
contextual information
continuous online optimization
contrastive divergence
converged algorithm
convergence
convert type
cookie ID
cookie matching
correct classification
correlation, difference from causation
cost function
cost of one impression.
See CPI.
cost per 1,000 impressions.
See CPM.
cost per click.
See CPC.
covariance_type parameter
CPC (cost per click), 2nd
CPI (cost of one impression)
CPM (cost per mille), 2nd
Criteo
CTR (click-through rate), 2nd
Cun, Yann Le
curse of dimensionality, 2nd
Cutter number
dac.tar.gz file
DAGs (directed acyclic graphs)
data array, 2nd
data axis, transforming
eigenvectors and eigenvalues
principal component analysis
data capture.
See capturing data on web.
Data class
data collection
managing at scale
naïve approach
data preparation
data reliability
data size
data types
data_categoricals array
data_non_categoricals array
datasets, large
decision tree algorithms.
See DT.
decisioning engine
click prediction for online advertising
contextual information
data preparation
decisioning engine model
feature engineering
information about placement
information about user
mapping predicted click-through rate to bid price
model training
decisions
deduplication
bolts
data
window
deep learning.
See also neural networks.
demand-side platform.
See DSP.
DESCR attribute
dimensional spaces
dimensionality
directed acyclic graphs.
See DAG.
distance
distance-based algorithms, 2nd
down-sampled dataset, 2nd, 3rd
drift
DSP (demand-side platform), 2nd, 3rd
DT (decision tree) algorithms
duplicate partitions
durability
eigenvalues
eigenvectors
EM (expectation maximization) algorithm, 2nd, 3rd
encoded data
energy-based learning
epsilon decreasing strategy
epsilon first strategy
epsilon greedy strategy
error attribute
Euclidean distance
evaluation
events
exchange, click prediction for online advertising and
ad monitoring
ad placement
bid win (or loss) notification
cookie matching
expectation maximization
explore-exploit conundrum
extensibility
f1 score
feature engineering
FeedForwardNetwork object
field grouping
filtering, user-based collaborative
fit method
flat geometries
flat reference structures
flexibility
FlumeJava
FNR (false negative rate)
forecasting
forward chaining
FP rate
FPR (false positive rate), 2nd
F-score, 2nd
FullConnection object
future applications of intelligent web
home healthcare
internet of things
personalized physical advertising
self-driving vehicle
semantic web
Gaussian mixture model.
See GMM.
Gaussian processes
GDBTs (gradient-boosted decision trees)
generalization
generalized linear models
Gerbrands
Gibbs measure
GLM (generalized linear model)
GMM (Gaussian mixture model), 2nd, 3rd
example of learning using
Gaussian distribution
general discussion
k-means and
gmm object
Google Glass
Google Now example
grayscale values
GUID (globally unique identifier)
Hadoop framework, Kafka plus
hand-built neuron
hashing trick
HDFS (Hadoop Distributed File System)
head tool
healthcare
hidden layer, 2nd
hidden nodes, 2nd, 3rd
hidden units, 2nd
hierarchical reference structures
higher-dimensional spaces
high-frequency trading
Hinton, Geoffrey
home healthcare
horizontal scaling
Huang, Jeubin
human-readable tags
HW (high watermark)
hyperbolic profile
hyperbolic tangent function
hyperplane
if-then clauses
incorrect classification
inference
inhibitory input
input layer, 2nd
input variables
instances
in-sync replicas
intelligence, evaluating
intelligent algorithms, 2nd, 3rd
classes of
artificial intelligence
machine learning
predictive analytics
data reliability and
data size and
evaluating performance of
evaluating intelligence
evaluating predictions
examples of
generalization and
Google Now example
inference and
lifecycle of
scaling characteristics of
what they are not
intelligent web
future applications of
home healthcare
internet of things
personalized physical advertising
self-driving vehicle
semantic web
social implications of
internet of things
invert_hash option
Iris dataset
IsFraud label
ISRs (in-sync replicas)
Kafka log-processing platform
consumer groups
design patterns
evaluating
plus Hadoop
plus Storm
replication in
message acknowledgement
replication, leaders, and in-sync replicas
KafkaClient object
keyword searches
KISS (keep it simple, stupid!)
k-means algorithm, 2nd, 3rd, 4th
kNN (k-nearest neighbors)
landing-page optimization
large datasets, classification with
latency
latent space
learning rate
learning vector quantization.
See LVQ.
leave-one-out technique
left-most neuron
lifecycle
of classifiers
of intelligent algorithms
linear approximation
linear models, 2nd
linear regression, 2nd, 3rd
linear relationship
linear supervised learning
link function
load method, 2nd
log shipping
log.dirs parameter
logging
logic inputs
logistic profile
logistic regression, 2nd, 3rd, 4th, 5th, 6th, 7th, 8th
log-odds, 2nd, 3rd, 4th
LVQ (learning vector quantization)
m1-1m dataset
MABs (multi-armed bandits)
epsilon decreasing strategy
epsilon first strategy
epsilon greedy strategy
extensions to
adversarial bandits
contextual bandits
MAC addresses
machine learning.
See ML.
make_ellipses method
mapping, predicted click-through rate to bid price
MapReduce
McCulloch, Warren
MCP model
mean_center
metaclassifier scheme
Millwheel
min_values
Minsky, Marvin
mirroring
missing values
ML (machine learning)
MLP (multilayer perceptron)
model training, 2nd, 3rd
modulo operator
MovieLens dataset
multi-armed bandits.
See MABs.
multiclass classification, 2nd, 3rd
multicluster operation
multilayer perceptrons
activation functions
backpropagation and
in scikit-learn
learned
multiple partitions
n dimensional space
n_components parameter
n_iter parameter
n-dimensional Gaussian
nearest-neighbor algorithms
negative classes
negative classification
negative exponential profile
negative sign
Netflix Prize example
neurons
new_data array
Ng, Andrew
NNs (neural networks)
multilayered
activation functions
backpropagation and
in scikit-learn
learned multilayered perceptron
overview
perceptrons
geometric interpretation of for two inputs
training
RBMs (Restricted Boltzmann Machines) and
BRBMs (Bernoulli Restricted Boltzmann Machines)
illustrative example
overview
noise
normal interval
nudge_dataset function
NumPy arrays, 2nd
odds of event
offset attribute
OMG (Object Management Group)
one-hot-encoding
online advertising.
See click prediction for online advertising.
online learning, 2nd
online optimization
only_unknowns=True parameter
ontologies
optimal performance
Orkut
orthogonal vectors
output error
output layer, 2nd
output node
output profiles
PA (predictive analytics), 2nd
Papert, Seymour
parametric model
PARC (Palo Alto Research Center)
partial derivative
partition attribute
PCA (principal component analysis), 2nd
PDFs (probability density functions), 2nd
perceptron learning algorithm
perceptrons
geometric interpretation of for two inputs
multilayer
activation functions
backpropagation and
in scikit-learn
learned
training
performance evaluation
evaluating intelligence
evaluating predictions
personalized physical advertising
petal length and width
physical advertising, personalized
pipe character
Pitts, Walter
placement, information about
plot_gmm_classifier
plotting output
positive classes
positive classification
positive values
post_normalize
pre_normalize
precision
predict_rating method
predictive analytics.
See PA.
PredictWallStreet
pre-prepared ads
principal component analysis.
See PCA.
prior knowledge
probabilities.out file
probability distribution functions.
See PDFs.
probability, obtaining with Gibbs measure
ProduceResponse objects
production rules
production stage
programmatic buying
publish-subscribe system
pull_handle method
PyBrain, 2nd
python-recsys package
–r switch
RatingCountMatrix class
RBM/LR pipeline
RBMs (Restricted Boltzmann Machines)
BRBMs (Bermoulli Restricted Boltzmann Machines)
illustrative example
overview
training
readable.models
recall, 2nd
receiver operating characteristic curve.
See ROC.
recommend method
recommender algorithm
recommending relevant content
distance and similarity
evaluating recommender
how recommender engines work
model-based recommendation using singular value decomposition
choosing movies for given user
choosing users for given movie
overview
Netflix Prize example
user-based collaborative filtering
regression algorithms
regret
accumulated
maximum
relevant content.
See recommending relevant content.
reliability of data
replication
in Kafka log-processing platform
message acknowledgement
replication, leaders, and in-sync replicas
Restricted Boltzmann Machines.
See RBMs.
retargeting
Rete algorithm
right-most neuron
RMSE (root-mean-square error)
ROC (receiver operating characteristic) curve, 2nd, 3rd
Rosenblatt, Frank
rows value
rudimentary ontology
rule-based algorithms
runtime performance
scalability
scale of probability
scaling characteristics, of intelligent algorithms
scikit-learn, 2nd, 3rd
SE (standard error)
second price auction
self-driving vehicle
semantic ontology
semantic web
semi-empirical approach
SGD (stochastic gradient descent)
similarity
similarity of probability
SimilarityMatrix class
SimpleConsumer
SimpleProducer, 2nd, 3rd
simplicity
single hidden units
single hyperplane
single perceptron
single test variable
single weight value
single-shot optimization
Single-Vector Lanczos Method
singular value decomposition.
See SVD.
size of data
sklearn.cluster.KMeans
sklearn.datasets.base.Bunch
sklearn.mixture.GMM
social implications of intelligent web
sort tool
sortModules method
specificity, 2nd
speed
square root profile
standard error.
See SE.
statistical algorithms, 2nd, 3rd, 4th
stochastic gradient descent.
See SGD.
structural algorithms, 2nd
structure
sum_probs_chosen method
SVD (singular value decom-position), 2nd, 3rd
choosing movies for given user
choosing users for given movie
overview
svd.compute method
svd.load_data method
svd.recommend method, 2nd
SVDLIBC library
SVMs (support vector machines)
target array, 2nd
target_names attribute
test topic
test_vw_file
testing stage
test-replicated-topic
tfidf value
threshold bias
TNR (true negative rate)
topic attribute
TPR (true positive rate), 2nd
train method
train_vw_file
trained model
trained neural networks, 2nd
training data
training pipeline
training stage
two-layer hidden network
two-layer perceptron
type I and II errors
ubiquitous computing
user-based collaborative filtering
userdict object
users, information about
utility problem
validation stage, 2nd
vehicles, self-driving
Vickrey auction
visible units, 2nd
volume
Vowpal Wabbit (VW), click prediction with
model calibration
preparing dataset
testing model
VW data format
Wald interval
web.
See intelligent web.
weighted sum, 2nd
weighted_ratings
weighting parameter
XOR function, 2nd, 3rd, 4th