Glossary · letter S
AI terms starting with S.
48 entries that start with S, each with its Nepali equivalent and a plain-language definition. Click any term to open its dedicated page.
-
Sample (Instance)
नमुना / इन्स्ट्यान्स
A single example or data point in a dataset.
-
Sampler
स्याम्पलर
The algorithm that guides how a diffusion model removes noise to form an image.
-
Sampling
नमुना लिने
Selecting a subset of data to represent and study a larger population.
-
Sampling Bias
नमुना पूर्वाग्रह
Error from collecting a sample that does not represent the whole population.
-
Scalability
मापनीयता
A system's ability to handle growing amounts of work or users without breaking.
-
Scene Understanding
दृश्य बोध
Interpreting the overall content and relationships of objects in an image or scene.
-
SDK
एसडीके
A software development kit providing tools and libraries to build on a platform.
-
Search Algorithm
खोज एल्गोरिदम
A method for exploring possible options to find a solution, such as a path or best move.
-
Seed
सिड
A starting number that makes a generative model's random output reproducible.
-
Self-Attention
स्व-एटेन्सन
A mechanism letting each part of a sequence weigh the importance of every other part.
-
Self-Supervised Learning
स्व-पर्यवेक्षित सिकाइ
A model creates its own labels from raw data, e.g. predicting missing words in a sentence.
-
Semantic Search
अर्थपूर्ण खोज
Searching by meaning rather than exact keywords, using embeddings.
-
Semantic Segmentation
अर्थपूर्ण खण्डीकरण
Labeling every pixel in an image with the class of object it belongs to.
-
Semi-Supervised Learning
अर्ध-पर्यवेक्षित सिकाइ
Learning from a small amount of labeled data combined with a large amount of unlabeled data.
-
Sensor
सेन्सर
A device that detects physical signals like light, sound, or motion for a system to use.
-
Sentiment Analysis
भावना विश्लेषण
Detecting whether text expresses positive, negative, or neutral emotion.
-
Sequence-to-Sequence
सिक्वेन्स-टु-सिक्वेन्स
A model that converts one sequence into another, such as a sentence into a translation.
-
Serverless
सर्भरलेस
A model where code runs on demand without managing the underlying servers.
-
SHAP
स्याप (SHAP)
A method that explains a model's prediction by attributing it to each feature.
-
Sharding
सार्डिङ
Splitting a large model or database across machines to scale beyond one device.
-
Sigmoid
सिग्मोइड
An activation function that squeezes values into a range between 0 and 1.
-
Skewness
विषमता
A measure of how lopsided or asymmetric a data distribution is.
-
SMOTE
स्मोट (SMOTE)
A technique that creates synthetic minority-class samples to fix class imbalance.
-
Softmax
सफ्टम्याक्स
A function that turns raw scores into probabilities that sum to one across classes.
-
Speech Recognition
वाक् पहिचान
Converting spoken language into written text.
-
Stacking
स्ट्याकिङ
Combining predictions from several models using another model on top.
-
Standard Deviation
मानक विचलन
A measure of how spread out values are around the mean.
-
Standardization
मानकीकरण
Rescaling features to have zero mean and unit variance for fair comparison.
-
State Space
अवस्था स्थान
The set of all possible situations a problem can be in, explored to find a solution.
-
Statistical Significance
तथ्याङ्कीय महत्त्व
When a result is unlikely to have occurred by random chance alone.
-
Stemming
स्टेमिङ
Cutting words down to a root form by chopping endings, e.g. 'running' to 'run'.
-
Stochastic Gradient Descent (SGD)
स्टोकास्टिक ग्र्याडियन्ट डिसेन्ट
Gradient descent that updates weights using small random batches for speed.
-
Stop Sequence
स्टप सिक्वेन्स
A specified string that tells a model to stop generating further text.
-
Stop Words
स्टप वर्ड्स
Common words like 'the' or 'and' often removed because they carry little meaning.
-
Streaming Data
स्ट्रिमिङ डाटा
Continuously generated data processed in real time as it arrives.
-
Stride
स्ट्राइड
How many pixels a convolution filter moves at each step across an image.
-
Strong AI
बलियो एआई
A theoretical AI with genuine consciousness and human-like understanding, not yet achieved.
-
Style Transfer
शैली स्थानान्तरण
Applying the visual style of one image to the content of another.
-
Subword
उप-शब्द
A fragment of a word used as a token so models can handle rare or new words.
-
Super-Resolution (Upscaling)
सुपर-रिजोल्युसन / अपस्केलिङ
Increasing an image's resolution and detail using AI.
-
Supervised Learning
पर्यवेक्षित सिकाइ
Training a model on labeled examples so it learns to map inputs to known outputs.
-
Support Vector Machine (SVM)
सपोर्ट भेक्टर मेसिन
An algorithm that separates data into classes by finding the best dividing boundary.
-
Sustainability (AI)
दिगोपन (एआई)
Reducing the energy use and environmental impact of training and running AI.
-
Swarm Intelligence
झुन्ड बुद्धिमत्ता
Collective problem-solving inspired by groups like ant colonies or bird flocks.
-
Symbolic AI
प्रतीकात्मक एआई
An older approach to AI based on explicit rules and logic rather than learning from data.
-
Synthetic Data
कृत्रिम डाटा
Artificially generated data used to train or test models when real data is limited.
-
Synthetic Media
कृत्रिम मिडिया
Images, audio, or video created or heavily altered by AI rather than captured directly.
-
System Prompt
सिस्टम प्रम्प्ट
A hidden instruction that sets a model's role, tone, and rules for a conversation.