Concept Glossary: O¶

This page lists English glossary entries for this letter. Entry bodies are generated from per-term source files with includes.

object¶

• Meaning: An object is an actual value that Python handles in memory. Numbers, strings, lists, DataFrames, and model instances can all be understood as objects, and each object carries data and behavior according to its type. In this sense, an object is not just a bundle of data; it is a runtime target that combines what kind of value it is with what it can do.
• Why it matters: Variable names may look similar, but the attributes and methods available depend on the actual object bound to the name. This helps readers interpret code by asking what kind of target is this? rather than only what is this name? It is also necessary for understanding Python behavior such as copying values, sharing references, and calling methods at the level of actual runtime objects.
• Related concepts: class, method, attribute
• Core Section: P2-8.6
• Appears in: P2-9.1, P2-11.1

object detection¶

• Meaning: Object detection is a vision task that predicts not only what is in an image, but also where each object is. It does not stop at there is a cat; it also asks where the cat is and how much space it occupies. The output is usually a set of category + location results for each detected object, not a single label.
• Why it matters: Object detection has a more complex output structure than image classification, and it is a representative example of deep learning expanding from category prediction to category and location prediction together. It helps separate classification, detection, and segmentation even when they all use image inputs. Understanding object detection also makes it easier to see why models such as YOLO are described as turning a multi-step vision pipeline into one prediction problem.
• Related concepts: bounding box, YOLO, image recognition, output structure
• Core Section: P1-9.2
• Appears in: P1-9.3, P1-10.1, P5-11.1, P6-19.2

objective function¶

• Meaning: An objective function is the overall criterion that a learning or optimization process tries to decrease or increase. It can combine sample-level loss, average loss over many samples, regularization penalties, and constraints into one rule for deciding what counts as better. A loss function can be a central component inside an objective function, but the objective function is the broader scoring rule.
• Why it matters: A model does not move toward being vaguely better; it is adjusted in the direction defined by the objective function. This concept connects loss functions, regularization, and constraints into the actual learning target. It also separates evaluation statements such as the accuracy is high from the quantity that training directly optimized.
• Related concepts: loss function, optimization, metric
• Core Section: P2-6.2
• Appears in: P2-6.3, P2-15.2

observation¶

• Meaning: An observation is new information or a result obtained after taking an action. In reinforcement learning it may be the next screen, number, or signal returned by the environment. In a service-style agent, it can be a search result, error message, file content, or test output used for the next decision.
• Why it matters: In multi-step problems, what is learned after an action can immediately change the next action or the decision to stop. Search results, error messages, tool outputs, and test results all update state and reshape the plan. This concept helps frame agent execution as a repeated cycle of action, observation, and selection rather than a single answer.
• Related concepts: agent, state, action
• Core Section: P1-14.3
• Appears in: P1-14.4, P1-14.5

operation¶

• Meaning: An operation is a task performed on a value or data structure, such as access, search, insertion, deletion, or traversal. Even with the same data, append to the end, insert in the middle, look up by name, and visit every item once are different operations. Data structures are chosen partly by which operations are expected to be common.
• Why it matters: The most natural data structure can change depending on which operations are frequent. This concept connects data structures not only to how data is shaped, but also to what work will be done and how often. It explains why arrays, hash-based structures, and graphs are useful in different situations.
• Related concepts: data structure, array, abstract data type
• Core Section: P2-9.1
• Appears in: P2-11.1, P2-12.1, P2-15.1

operation¶

• Meaning: 서비스를 한 번 실행하는 데서 끝내지 않고, 반복 사용 속에서 비용, 오류, 속도, 품질을 계속 관찰하고 조정하는 일입니다. 모델 성능을 확인하는 것만이 아니라, 실패가 어디서 나고 어떤 제한 때문에 병목이 생기며 어떤 기록을 남겨야 하는지까지 함께 다루는 지속 관리 층에 가깝습니다. 즉 개발이 만들기라면 운영은 계속 굴러가게 만들기에 더 가까운 문제입니다.
• Why it matters: AI 서비스는 모델 품질만 맞는다고 끝나지 않고, 실패 대응과 예산, 사용 경험을 함께 관리해야 오래 유지할 수 있기 때문입니다. 이 개념이 있어야 좋은 모델을 골랐다와 서비스가 안정적으로 굴러간다를 같은 문제로 보지 않게 되고, 레이트 리밋, 재시도, 로그, 평가를 모두 운영 판단의 한 묶음으로 읽게 됩니다. 또한 운영을 이해해야 같은 성능 수치라도 비용이 과도하거나 재시도가 잦으면 실제 서비스 품질은 나빠질 수 있다는 점도 더 구체적으로 보게 됩니다.
• Related concepts: cost, latency, harness, retry
• Core Section: P1-14.6
• Appears in: P1-14.2, P1-14.5

optimization¶

• Meaning: Optimization is the problem of finding a better value among many candidates while considering criteria and constraints. It does not always mean computing one perfect mathematical answer. It often refers to the whole process of adjusting values in a direction that improves the current objective function.
• Why it matters: Optimization lets readers understand learning as a search for values that reduce loss, not as memorizing answers. It gives a shared frame for model training, path search, and resource allocation: what counts as good, under what constraints, and which values are being adjusted. It also explains why learning is usually expressed as an iterative process.
• Related concepts: objective function, gradient descent, constraint
• Core Section: P2-6.1
• Appears in: P2-6.2, P2-6.3, P2-15.2

optimizer¶

• Meaning: An optimizer is the rule or procedure that decides how to update parameters using gradients computed by backpropagation. If a gradient indicates which direction reduces loss, the optimizer decides how far and in what manner to move in that direction. Training therefore needs an update rule in addition to a loss function and gradients.
• Why it matters: Loss and gradients alone do not complete learning, because the step size and update style strongly affect training speed and stability. This concept separates which direction should we go? from how should we move? It also explains why the same model can train differently under update rules such as SGD and Adam.
• Related concepts: gradient, learning rate, loss function
• Core Section: P5-7.1
• Appears in: P5-7.2, P5-7.3

orchestration¶

• Meaning: Orchestration is the coordination layer that connects models, data, tools, and apps into one workflow under specific orders and conditions. It decides what to check first, when to search, when to call a tool, when to stop, and how to move to the next step after failure.
• Why it matters: Real service quality often depends less on one model's raw ability than on when components are connected and where the workflow stops. The same model can produce different cost, speed, safety, and failure patterns under different orchestration. This concept separates model performance problems from workflow design problems.
• Related concepts: application, agent, Model Context Protocol, MCP, tool use, stop condition
• Core Section: P1-14.1
• Appears in: P1-14.3

out-of-vocabulary, OOV¶

• Meaning: Out-of-vocabulary, or OOV, refers to a word or token that falls outside the current vocabulary or tokenization rules. A human may see a meaningful expression, but the model may not be able to represent it as familiar units. OOV is therefore a signal that an input expression is outside the current vocabulary system, not simply that the sentence is strange.
• Why it matters: When many OOV items appear in classification or search, the model may effectively read less of the input. This helps interpret poor performance as a possible tokenizer or vocabulary design issue, not only as a model capability issue. Domain terms, new words, and product codes can be clear to people while still being fragmented into unfamiliar pieces for a model.
• Related concepts: tokenization, token coverage, embedding
• Core Section: P7-4.2
• Appears in: P7-4.1, P7-4.3, P7-summary

outcome¶

• Meaning: An outcome is one individual result that can occur in a single trial. In a coin toss, heads and tails are outcomes; in a die roll, values such as 1, 2, and 3 are outcomes. It is the smallest individual case used to build a probability problem.
• Why it matters: Outcomes are the basic unit for distinguishing events from the sample space and for clarifying what probability is assigned to. This concept prevents confusing an event such as an even number appears with the individual outcomes that make up that event. It also shows that defining the outcome is part of defining the probability problem itself.
• Related concepts: event, sample space, probability
• Core Section: P2-5.1
• Appears in: P2-5.2, P2-5.3, P2-6.1

outlier¶

• Meaning: An outlier is a value that appears unusually far from the overall pattern of values. Being visibly different does not automatically mean the value is wrong. It may be an input mistake, a rare real event, or an important warning signal.
• Why it matters: Outliers can come from input errors, measurement problems, or rare but meaningful cases, so they must be interpreted before being discarded. This concept makes readers consider how extreme values affect interpretation and model training, not only the average pattern near the center of a distribution. In small datasets, only a few outliers can strongly shift means, variances, and regression lines.
• Related concepts: distribution, plot, histogram
• Core Section: P2-13.1
• Appears in: P2-13.2

output¶

• Meaning: An output is the result a model is expected to produce. It may be a number, a category, a sentence, a report, or a candidate list. The key point is that output is not merely the visible final answer; it is also a design choice that reveals what form the problem has been given.
• Why it matters: The same real-world situation can become very different modeling tasks depending on the output definition. Defining a problem is closely tied to deciding what counts as input and what should be returned as output. This concept explains why the same data can lead to classification, regression, summarization, recommendation, or comparison-report generation, each with different evaluation criteria.
• Related concepts: input, label, model, output structure, task definition
• Core Section: P1-4.2
• Appears in: P1-8.1, P1-10.1, P1-14.1, P1-14.3, P1-14.5, P1-15.3

output format¶

• Meaning: Output format is the condition that specifies the shape of the returned answer, such as a table, list, JSON object, paragraph, or step-by-step explanation. It is an interface condition that is separate from the core content itself. It decides the wrapper in which the result should be delivered.
• Why it matters: The same content can be much easier or harder to post-process, review, and reuse depending on its output format. This concept separates what should be answered? from in what structure should it be returned? It also explains why a useful answer can still break an automation pipeline or review process if the format is wrong.
• Related concepts: prompt, constraint, instruction
• Core Section: P1-12.1

output structure¶

• Meaning: Output structure is the designed result frame that determines how a computation should be delivered, such as a comparison report, review candidate queue, or target label candidates. It asks not only what is the answer?, but also what container should hold the result, and who will use it next? Choosing between a single number, a human-readable comparison table, or a candidate set for later training is a choice of output structure.
• Why it matters: The same source data leads to different dataset designs depending on whether the result becomes a comparison table, a review workflow, or future training candidates. Output structure must be decided before review procedures and automation steps can be designed. This concept separates what answer should be produced? from what workflow-ready form should carry that answer?
• Related concepts: comparison report, target, sample, task definition
• Core Section: P3-2.2
• Appears in: P3-index, P3-1.1, P3-1.2, P3-1.3, P3-2.1, P3-3.2, P3-4.3, P3-8.2, P3-9.1, P3-9.2, P3-9.3, P3-9.5, P3-summary, P7-index, P7-1.1, P7-2.1, P7-5.1, P7-summary

overfitting¶

• Meaning: Overfitting is the condition in which a model fits the training data too closely and performs worse on new data. The model has learned not only real repeated patterns, but also accidental fluctuations or noise that appeared only in the training set.
• Why it matters: Overfitting is central to understanding why learning from data is not the same as memorization. It explains why training scores and validation scores are checked separately, and why regularization and early stopping become important as models grow more complex. It also separates high training performance from useful performance on new cases.
• Related concepts: generalization, underfitting, regularization
• Core Section: P4-5.1
• Appears in: P1-3.2, P4-5.2, P4-14.2, P4-16.2, P5-8.1, P5-8.2