WBCHSE Applied AI flashcards

32 cards. Read the term and the definition here. The deck at the top of this page is optional.

Artificial intelligence

Artificial intelligence is the field of building systems that perform tasks associated with human intelligence, such as perception, reasoning, learning, planning, and decision-making.

Turing Test

The Turing Test, proposed by Alan Turing in 1950, judges a machine as intelligent if its text responses are indistinguishable from a human's in conversation.

Narrow AI

Narrow AI performs a specific task, such as translation or recommendation, while general AI would match humans across many tasks; current deployed systems are narrow.

Algorithmic bias

Algorithmic bias is systematic unfairness in AI outputs caused by unrepresentative data, flawed labels, or design choices, often disadvantaging a group without any intent.

Data minimisation

Data minimisation is the privacy principle of collecting and retaining only the personal data needed for a stated purpose, reducing exposure if a breach occurs.

Human-in-the-loop

Human-in-the-loop means a person reviews or approves an AI system's decision before it affects someone, and is used for high-stakes credit, health, or education decisions.

Mutable versus immutable

A mutable object, such as a list, can be changed after creation; an immutable object, such as a tuple or string, cannot, so operations return a new object.

Python floor division

Floor division uses the // operator and returns the quotient rounded down to the nearest integer; for positive numbers it discards the remainder.

Zero-based indexing

Zero-based indexing means the first element of a Python sequence has index 0, the second has index 1, and negative indices count backward from the end.

ValueError

A ValueError occurs when a function receives an argument of the right type but an invalid value, such as int('abc') or unpacking the wrong number of values.

Median

The median is the middle value of a dataset sorted in order; with an even number of values it is the mean of the two middle values and is robust to outliers.

Histogram

A histogram groups numeric data into bins and plots frequency or density per bin, showing the shape, centre, spread, and possible outliers of a distribution.

Correlation coefficient

The Pearson correlation coefficient r ranges from -1 to 1, measuring the strength and direction of a linear association between two numeric variables; zero means no linear association.

Data cleaning

Data cleaning detects and fixes missing values, duplicates, inconsistent formats, invalid types, and outliers so that analysis and models use reliable input.

Supervised learning

Supervised learning trains a model on labelled examples, pairing input features with a known target, to predict labels or values for new unseen inputs.

Overfitting

Overfitting occurs when a model learns training-data noise and specific details so well that it performs poorly on new data; more data and regularization can reduce it.

Confusion matrix

A confusion matrix tabulates true positives, false positives, false negatives, and true negatives, allowing precision, recall, and F1 to be calculated for a classifier.

Data leakage

Data leakage happens when information from the target or test set enters training features or preprocessing, producing evaluation scores that are too optimistic.

Tokenisation

Tokenisation splits text into smaller units such as words, subwords, or characters, which are then converted into numbers for an NLP model.

TF-IDF

TF-IDF scores a term by its frequency in a document multiplied by inverse document frequency, so terms common across many documents receive lower weight.

CNN

A convolutional neural network uses convolution filters to learn local patterns in images, usually followed by pooling and fully connected layers for classification.

Object detection

Object detection locates multiple objects in an image by outputting bounding boxes and class labels, unlike classification, which assigns one label to the whole image.

Predictive maintenance

Predictive maintenance uses sensor data and models to estimate when equipment will fail, so maintenance can be scheduled before a breakdown instead of at fixed intervals.

Recommendation system

A recommendation system ranks items for a user by learning from past behaviour, item similarity, or both, and is common in retail and media platforms.

Precision agriculture

Precision agriculture uses sensor, satellite, and image data to target water, fertiliser, and pest control to specific field zones, reducing waste and improving yield.

Jupyter Notebook

A Jupyter Notebook is an interactive document of code, output, and text cells, widely used for data exploration and sharing reproducible analysis.

Large language model

A large language model is a neural network trained on massive text to predict and generate language; it can produce fluent but sometimes false or hallucinated content.

API

An API, or application programming interface, lets one program request data or services from another through defined inputs and outputs, often exchanging JSON over HTTP.

Data dictionary

A data dictionary documents each dataset field's name, type, units, allowed range, and missing-value meaning, helping others use the data correctly.

Reproducibility

Reproducibility means another person can rerun a project and obtain the same results, supported by fixed random seeds, pinned library versions, and documented steps.

Minimum viable product

A minimum viable product is the smallest version of a product that lets real users test its core value proposition, enabling learning before large investment.

Value proposition

A value proposition states the customer segment, the problem being solved, and why the product is better than alternatives, guiding product and pitch decisions.