{
  "notice": "bxzxxr MCP archive: a record of what public MCP servers declared when we listed them (we never call their tools). It is data, not a rating: nothing here says a server is safe, secure, verified or trustworthy, and a server can change after we looked. Secrets we detected are redacted; redacted tool lists are withheld and only their hash is shown.",
  "infrastructure": {
    "host": "dictionaryofml.org",
    "site": "dictionaryofml.org",
    "addresses": [
      {
        "address": "178.105.197.122",
        "endpoints": 1,
        "shared_infrastructure": false,
        "network": {
          "asn": 24940,
          "name": "HETZNER-AS - Hetzner Online GmbH",
          "prefix": "178.104.0.0/15",
          "country": "DE"
        }
      }
    ],
    "addresses_recorded": true,
    "address_history": [
      {
        "addresses": [
          "178.105.197.122"
        ],
        "first_seen": "2026-11-21T19:06:01.995Z",
        "last_seen": "2026-11-21T19:06:01.995Z",
        "checks": 1
      }
    ],
    "hosting": [],
    "software": [
      {
        "value": "nginx",
        "evidence": "Server header \"nginx/1.28.3 (Ubuntu)\""
      }
    ],
    "server_info": {
      "name": "dictionaryofml",
      "version": "1.1.0",
      "protocol_version": "2025-06-18"
    },
    "headers": {
      "server": "nginx/1.28.3 (Ubuntu)",
      "strict-transport-security": "max-age=31536000; includeSubDomains"
    },
    "auth": null
  },
  "related": {
    "same_address": {
      "total": 0,
      "sample": []
    },
    "same_site": {
      "total": 0,
      "sample": []
    },
    "same_tool_list": {
      "total": 0,
      "sample": []
    },
    "same_publisher": {
      "publisher": "org.dictionaryofml",
      "total": 0,
      "sample": []
    },
    "same_repository": {
      "repository_url": "https://github.com/alexjungaalto/dictionaryofml",
      "total": 0,
      "sample": []
    },
    "note": "Shared addresses on big platforms (Cloudflare, Vercel and the like) put unrelated servers together; those are marked shared_infrastructure. A match is a pivot, not a link between operators."
  },
  "endpoint": {
    "id": "45178",
    "server_name": "org.dictionaryofml/dictionary",
    "url": "https://dictionaryofml.org/mcp",
    "host": "dictionaryofml.org",
    "transport": "streamable-http",
    "last_checked_at": "2026-11-21T19:06:01.995Z",
    "last_state": "listed",
    "last_detail": null,
    "consecutive_failures": 0,
    "current_sha": "643cf6182b6370bd5a8cdb82792808c065a0ba3290ac610ec9f94ecdcc647dd9",
    "first_listed_at": "2026-11-21T19:06:01.995Z",
    "created_at": "2026-11-21T13:10:02.709Z",
    "resolved_addresses": [
      "178.105.197.122"
    ]
  },
  "server": {
    "name": "org.dictionaryofml/dictionary",
    "title": "Dictionary of Applied Machine Learning",
    "description": "Look up, search and cross-reference machine-learning terms from dictionaryofml.org.",
    "version": "1.1.0",
    "registry_status": "active",
    "repository_url": "https://github.com/alexjungaalto/dictionaryofml",
    "website_url": "https://dictionaryofml.org",
    "published_at": "2026-11-09T11:32:43.520Z",
    "updated_at": "2026-11-09T11:32:43.520Z",
    "first_seen": "2026-11-21T13:10:01.949Z"
  },
  "other_endpoints": [],
  "intel": {
    "capabilities": [],
    "evm_addresses": [],
    "chains": [],
    "tokens": [],
    "computed_at": "2026-11-22T15:49:08.804Z"
  },
  "operator": {
    "key": "dictionaryofml.org",
    "name": "dictionaryofml.org",
    "platform": false
  },
  "needs_sign_in": false,
  "reachability": {
    "window_days": 30,
    "checks": 1,
    "answered": 1,
    "answered_percent": 100,
    "avg_listing_latency_ms": 1458,
    "first_check_in_window": "2026-11-21T19:06:01.995Z",
    "consecutive_failures_now": 0,
    "days": [
      {
        "day": "2026-11-21",
        "checks": 1,
        "answered": 1,
        "last_state": "listed"
      }
    ]
  },
  "tool_list": {
    "sha256": "643cf6182b6370bd5a8cdb82792808c065a0ba3290ac610ec9f94ecdcc647dd9",
    "tools": 6,
    "resources": 155,
    "prompts": 0,
    "first_seen": "2026-11-21T19:06:03.455Z",
    "tool_names": [
      "list_all_terms",
      "lookup_term",
      "search_terms",
      "get_related_terms",
      "list_learning_paths",
      "get_learning_path"
    ],
    "declared": {
      "tools": [
        {
          "name": "list_all_terms",
          "title": "List every term",
          "description": "List every term of the Dictionary of Applied Machine Learning, one per line, as its display name followed by its key. Takes no arguments. Use it to find the key for a term before calling one of the tools that takes one.",
          "inputSchema": {
            "type": "object",
            "properties": {},
            "additionalProperties": false
          },
          "annotations": {
            "readOnlyHint": true,
            "openWorldHint": false,
            "idempotentHint": true,
            "destructiveHint": false
          }
        },
        {
          "name": "lookup_term",
          "title": "Read one entry",
          "description": "The full dictionary entry for one term: its abstract, its definition, the works it cites and its cross-references. Use it when a specific term is already known; use search_terms when it is not.",
          "inputSchema": {
            "type": "object",
            "required": [
              "term"
            ],
            "properties": {
              "term": {
                "type": "string",
                "description": "A term, given either as its key (the lowercase one-word identifier, e.g. 'erm', 'gradientdescent') or as its display name (e.g. 'empirical risk minimization'). Matching is case-insensitive and tolerates small differences in spelling; an input that resembles no published term returns a message saying so rather than a wrong entry."
              }
            },
            "additionalProperties": false
          },
          "annotations": {
            "readOnlyHint": true,
            "openWorldHint": false,
            "idempotentHint": true,
            "destructiveHint": false
          }
        },
        {
          "name": "search_terms",
          "title": "Search the dictionary",
          "description": "Search the term names, abstracts and definitions for a keyword or phrase and return the best matches, each with its abstract. Use it when the right term is not known in advance.",
          "inputSchema": {
            "type": "object",
            "required": [
              "query"
            ],
            "properties": {
              "limit": {
                "type": "integer",
                "default": 8,
                "maximum": 50,
                "minimum": 1,
                "description": "How many matches to return, best first. Defaults to 8."
              },
              "query": {
                "type": "string",
                "minLength": 1,
                "description": "The words to search for, e.g. 'overfitting' or 'gradient step size'. Matching is case-insensitive; a term whose name contains the query ranks above one where only the definition does."
              }
            },
            "additionalProperties": false
          },
          "annotations": {
            "readOnlyHint": true,
            "openWorldHint": false,
            "idempotentHint": true,
            "destructiveHint": false
          }
        },
        {
          "name": "get_related_terms",
          "title": "Follow the cross-references",
          "description": "The terms one entry cross-references, its See-also list, each with its abstract. Use it to widen a topic by one step from a term already in hand.",
          "inputSchema": {
            "type": "object",
            "required": [
              "term"
            ],
            "properties": {
              "term": {
                "type": "string",
                "description": "A term, given either as its key (the lowercase one-word identifier, e.g. 'erm', 'gradientdescent') or as its display name (e.g. 'empirical risk minimization'). Matching is case-insensitive and tolerates small differences in spelling; an input that resembles no published term returns a message saying so rather than a wrong entry."
              }
            },
            "additionalProperties": false
          },
          "annotations": {
            "readOnlyHint": true,
            "openWorldHint": false,
            "idempotentHint": true,
            "destructiveHint": false
          }
        },
        {
          "name": "list_learning_paths",
          "title": "List the reading orders",
          "description": "The author's curated reading orders through the dictionary: one per topic (how ChatGPT works, AI foundations, explainable and trustworthy AI, mathematics of AI, ...), each with its audience and its terms in the order to read them. Takes no arguments. Use when someone asks what to learn, read, or teach about a topic, or how to structure a course or seminar, so the answer follows the author's order instead of an improvised one.",
          "inputSchema": {
            "type": "object",
            "properties": {},
            "additionalProperties": false
          },
          "annotations": {
            "readOnlyHint": true,
            "openWorldHint": false,
            "idempotentHint": true,
            "destructiveHint": false
          }
        },
        {
          "name": "get_learning_path",
          "title": "Read one reading order",
          "description": "One reading order with the abstract of every term in it, in the order to read them. Give either `topic` or `terms`; giving neither returns the list of available reading orders.",
          "inputSchema": {
            "type": "object",
            "properties": {
              "terms": {
                "type": "array",
                "items": {
                  "type": "string"
                },
                "minItems": 1,
                "description": "An explicit reading order instead of a curated one: term keys or display names, kept in the given order. This is the form a selection on alexjung.at/baukasten/ takes, whose URL fragment reads '#terms=llm,transformer'."
              },
              "topic": {
                "type": "string",
                "minLength": 1,
                "description": "Which reading order to return: a path id (e.g. 'llm', 'grundlagen', 'mathematik') or words from its title or its audience note. Matching is case-insensitive."
              }
            },
            "additionalProperties": false
          },
          "annotations": {
            "readOnlyHint": true,
            "openWorldHint": false,
            "idempotentHint": true,
            "destructiveHint": false
          }
        }
      ],
      "tools_omitted": 0,
      "resources": [
        {
          "name": "Reading order: Wie funktioniert ChatGPT? (LLMs)",
          "uri": "dictml://path/llm",
          "mimeType": "text/markdown",
          "description": "LLMs · für alle, Öffentlichkeit & Firmen — 3 terms"
        },
        {
          "name": "Reading order: KI-Grundlagen",
          "uri": "dictml://path/grundlagen",
          "mimeType": "text/markdown",
          "description": "ML Core · für Bildungswerke, AK, Einsteiger — 96 terms"
        },
        {
          "name": "Reading order: Erklärbare & vertrauenswürdige KI",
          "uri": "dictml://path/xai",
          "mimeType": "text/markdown",
          "description": "XAI · für Management, Compliance, Betriebsräte — 8 terms"
        },
        {
          "name": "Reading order: KI, Recht & AI Act",
          "uri": "dictml://path/recht",
          "mimeType": "text/markdown",
          "description": "für AK, Betriebsräte, Datenschutzbeauftragte — 1 terms"
        },
        {
          "name": "Reading order: Datenschutz & Federated Learning",
          "uri": "dictml://path/fl",
          "mimeType": "text/markdown",
          "description": "für datensensible Betriebe und die öffentliche Hand — 1 terms"
        },
        {
          "name": "Reading order: KI in der Praxis (ML-Systeme)",
          "uri": "dictml://path/systeme",
          "mimeType": "text/markdown",
          "description": "für Unternehmen und Technik-Teams — 4 terms"
        },
        {
          "name": "Reading order: Reinforcement Learning",
          "uri": "dictml://path/rl",
          "mimeType": "text/markdown",
          "description": "RL · für Technik-Teams, Robotik, Logistik, Spieleentwicklung — 3 terms"
        },
        {
          "name": "Reading order: Mathematik der KI",
          "uri": "dictml://path/mathematik",
          "mimeType": "text/markdown",
          "description": "für Technik-Teams, Aktuar:innen, Fortgeschrittene — 31 terms"
        },
        {
          "name": "Laplacian matrix",
          "uri": "dictml://term/LapMat",
          "mimeType": "text/markdown",
          "description": "The Laplacian matrix of a weighted graph with n nodes is the n × n matrix L(G) = D(G) - A(G), the degree matrix minus the adjacency matrix of the edge weights"
        },
        {
          "name": "accuracy",
          "uri": "dictml://term/accuracy",
          "mimeType": "text/markdown",
          "description": "Accuracy is the fraction of correct predictions made by a hypothesis on a dataset with a finite label space, equal to one minus the average 0/1 loss"
        },
        {
          "name": "artificial neural network (ANN)",
          "uri": "dictml://term/ann",
          "mimeType": "text/markdown",
          "description": "An artificial neural network (ANN) is a hypothesis space whose hypothesis maps are all built on one directed graph"
        },
        {
          "name": "attention",
          "uri": "dictml://term/attention",
          "mimeType": "text/markdown",
          "description": "Attention is a mechanism that models the dependencies between the tokens that make up a data point, such as the words in a sentence or the pixel patches of an image"
        },
        {
          "name": "autoencoder",
          "uri": "dictml://term/autoencoder",
          "mimeType": "text/markdown",
          "description": "An autoencoder is a machine learning method that learns an encoder and a decoder together"
        },
        {
          "name": "backpropagation",
          "uri": "dictml://term/backpropagation",
          "mimeType": "text/markdown",
          "description": "Backpropagation is an algorithm that computes the gradient of the loss with respect to the model parameters of an artificial neural network"
        },
        {
          "name": "bagging",
          "uri": "dictml://term/bagging",
          "mimeType": "text/markdown",
          "description": "Bagging is an ensemble method that uses randomly sampled subsets of a dataset as training sets for its base learners"
        },
        {
          "name": "base learner",
          "uri": "dictml://term/baselearner",
          "mimeType": "text/markdown",
          "description": "A base learner is the machine learning method that an ensemble method runs several times: an algorithm that reads a training set and returns a learned hypothesis"
        },
        {
          "name": "baseline",
          "uri": "dictml://term/baseline",
          "mimeType": "text/markdown",
          "description": "A baseline is a reference level against which the performance of a trained model is compared"
        },
        {
          "name": "binary classification",
          "uri": "dictml://term/binclass",
          "mimeType": "text/markdown",
          "description": "Binary classification refers to machine learning applications with a binary label space"
        },
        {
          "name": "boosting",
          "uri": "dictml://term/boosting",
          "mimeType": "text/markdown",
          "description": "Boosting is an ensemble method that sequentially combines simple base learners, also called weak learners"
        },
        {
          "name": "bootstrap",
          "uri": "dictml://term/bootstrap",
          "mimeType": "text/markdown",
          "description": "The bootstrap is a stochastic algorithm that generates perturbed variants, referred to as bootstrap samples, of a given dataset"
        },
        {
          "name": "concept activation vector (CAV)",
          "uri": "dictml://term/cav",
          "mimeType": "text/markdown",
          "description": "A concept activation vector (CAV) is a direction in the activations of one hidden layer of a deep net that stands for a concept named by a user"
        },
        {
          "name": "classification",
          "uri": "dictml://term/classification",
          "mimeType": "text/markdown",
          "description": "Classification is the task of predicting a discrete-valued label for a given data point, based solely on its features"
        },
        {
          "name": "cluster",
          "uri": "dictml://term/cluster",
          "mimeType": "text/markdown",
          "description": "A cluster is a subset of the data points of a dataset that are more similar to each other than to the data points outside the subset"
        },
        {
          "name": "cluster centroid",
          "uri": "dictml://term/clustercentroid",
          "mimeType": "text/markdown",
          "description": "A cluster centroid is a vector that represents a cluster of data points: among all vectors, it minimizes the summed loss over the data points of the cluster, where the loss measures how poorly a vecto"
        },
        {
          "name": "clustering",
          "uri": "dictml://term/clustering",
          "mimeType": "text/markdown",
          "description": "Clustering decomposes a dataset into subsets, called clusters, such that data points within the same cluster are more similar to each other than to data points in other clusters"
        },
        {
          "name": "clustering error",
          "uri": "dictml://term/clusteringerror",
          "mimeType": "text/markdown",
          "description": "The clustering error of a clustering is the average, over the data points of a dataset, of a loss that measures how poorly a data point is represented by the cluster it is assigned to"
        },
        {
          "name": "confusion matrix",
          "uri": "dictml://term/cm",
          "mimeType": "text/markdown",
          "description": "The confusion matrix of a hypothesis on a finite dataset with K label values is the K × K matrix whose entry in row y and column ŷ counts the data points with true label y and prediction ŷ"
        },
        {
          "name": "convex",
          "uri": "dictml://term/convex",
          "mimeType": "text/markdown",
          "description": "A subset of the Euclidean space is convex if it contains the line segment between any two of its points"
        },
        {
          "name": "correlation",
          "uri": "dictml://term/correlation",
          "mimeType": "text/markdown",
          "description": "Correlation measures the strength and direction of the linear relationship between two real-valued RVs"
        },
        {
          "name": "covariance matrix",
          "uri": "dictml://term/covmtx",
          "mimeType": "text/markdown",
          "description": "The covariance matrix of a random vector consists of the variances of the individual entries on its main diagonal and the covariances between different entries off the diagonal"
        },
        {
          "name": "data",
          "uri": "dictml://term/data",
          "mimeType": "text/markdown",
          "description": "Data are representations of information, recorded in a form suitable for storage, communication and processing, and the word carries three technical senses in machine learning"
        },
        {
          "name": "data augmentation",
          "uri": "dictml://term/dataaug",
          "mimeType": "text/markdown",
          "description": "Data augmentation enlarges a training set with synthetic data points obtained from transformations or perturbations of existing ones"
        },
        {
          "name": "data imputation",
          "uri": "dictml://term/dataimputation",
          "mimeType": "text/markdown",
          "description": "Data imputation replaces missing attribute values of data points by estimates computed from the attribute values that were recorded"
        },
        {
          "name": "data leakage",
          "uri": "dictml://term/dataleakage",
          "mimeType": "text/markdown",
          "description": "Data leakage arises when a machine learning method uses features that are available during training but not at prediction time"
        },
        {
          "name": "data point",
          "uri": "dictml://term/datapoint",
          "mimeType": "text/markdown",
          "description": "A data point is the elementary information-carrying unit on which machine learning methods operate"
        },
        {
          "name": "dataset",
          "uri": "dictml://term/dataset",
          "mimeType": "text/markdown",
          "description": "A dataset is a finite collection of data points on which model training and validation are performed"
        },
        {
          "name": "density-based spatial clustering of applications with noise (DBSCAN)",
          "uri": "dictml://term/dbscan",
          "mimeType": "text/markdown",
          "description": "Density-based spatial clustering of applications with noise (DBSCAN) is a clustering method for data points with numeric feature vectors that finds clusters of any shape and decides their number itsel"
        },
        {
          "name": "decision boundary",
          "uri": "dictml://term/decisionboundary",
          "mimeType": "text/markdown",
          "description": "The decision boundary of a classifier consists of the feature vectors that lie arbitrarily close to two or more decision regions"
        },
        {
          "name": "decision region",
          "uri": "dictml://term/decisionregion",
          "mimeType": "text/markdown",
          "description": "A decision region of a classifier is the set of all feature vectors that the classifier maps to one particular label value"
        },
        {
          "name": "decision tree",
          "uri": "dictml://term/decisiontree",
          "mimeType": "text/markdown",
          "description": "A decision tree is a flowchart-like representation of a hypothesis map: a root node receives the feature vector of a data point and evaluates a basic test on it"
        },
        {
          "name": "decoder",
          "uri": "dictml://term/decoder",
          "mimeType": "text/markdown",
          "description": "A decoder is a map that turns the code of a data point, the short vector an encoder computes from it, back into the space the data point itself lives in"
        },
        {
          "name": "deep net",
          "uri": "dictml://term/deepnet",
          "mimeType": "text/markdown",
          "description": "A deep net is an artificial neural network whose hypothesis maps are compositions of several layers, each of which applies an affine map and then an activation function to every entry of the result"
        },
        {
          "name": "dimensionality reduction",
          "uri": "dictml://term/dimred",
          "mimeType": "text/markdown",
          "description": "Dimensionality reduction uses a map Φ: Rd → Rd' to replace a raw feature vector x = (x1, …, xd)ᵀ by a shorter feature vector z = (z1, …, zd')ᵀ, with d' < d"
        },
        {
          "name": "distribution shift",
          "uri": "dictml://term/distshift",
          "mimeType": "text/markdown",
          "description": "A data point carries features and a label, drawn together from one joint probability distribution"
        },
        {
          "name": "degree of belonging",
          "uri": "dictml://term/dob",
          "mimeType": "text/markdown",
          "description": "A degree of belonging is a number that indicates the extent to which a data point belongs to a cluster, one for each pair of a data point and a cluster of a clustering"
        },
        {
          "name": "edge weight",
          "uri": "dictml://term/edgeweight",
          "mimeType": "text/markdown",
          "description": "An edge weight is a nonnegative number Ai,i' attached to the edge i,i' of a weighted graph that measures how strongly the two nodes it joins are connected"
        },
        {
          "name": "effective dimension",
          "uri": "dictml://term/effdim",
          "mimeType": "text/markdown",
          "description": "The effective dimension of a hypothesis space is its size as seen through the probability distribution of the features: the number of model parameters that make a difference on the data points the dis"
        },
        {
          "name": "eigenvalue",
          "uri": "dictml://term/eigenvalue",
          "mimeType": "text/markdown",
          "description": "An eigenvalue of a square matrix A is a number λ for which some nonzero vector x satisfies Ax = λx"
        },
        {
          "name": "eigenvector",
          "uri": "dictml://term/eigenvector",
          "mimeType": "text/markdown",
          "description": "An eigenvector of a square matrix A is a nonzero vector x that A maps to a multiple of itself, A x = λ x"
        },
        {
          "name": "expectation–maximization (EM)",
          "uri": "dictml://term/em",
          "mimeType": "text/markdown",
          "description": "The expectation–maximization (EM) algorithm is an iterative optimization method for a particular class of maximum likelihood problems"
        },
        {
          "name": "encoder",
          "uri": "dictml://term/encoder",
          "mimeType": "text/markdown",
          "description": "An encoder is a map that sends the raw feature vector of a data point to a shorter vector, a code of d' numbers with d' below the number d of raw features"
        },
        {
          "name": "ensemble",
          "uri": "dictml://term/ensemble",
          "mimeType": "text/markdown",
          "description": "An ensemble method combines several machine learning methods, each referred to as a base learner"
        },
        {
          "name": "empirical risk minimization (ERM)",
          "uri": "dictml://term/erm",
          "mimeType": "text/markdown",
          "description": "Empirical risk minimization (ERM) selects a hypothesis by minimizing the average loss on a given training set D(train), the empirical risk"
        },
        {
          "name": "eigenvalue decomposition (EVD)",
          "uri": "dictml://term/evd",
          "mimeType": "text/markdown",
          "description": "An eigenvalue decomposition (EVD) of a square matrix breaks the linear map it represents into elementary operations: expanding a vector in the eigenvectors of the matrix, multiplying each expansion co"
        },
        {
          "name": "explainability",
          "uri": "dictml://term/explainability",
          "mimeType": "text/markdown",
          "description": "Explainability is the property of a machine learning method whose explanations let a human user anticipate the predictions it delivers"
        },
        {
          "name": "explanation",
          "uri": "dictml://term/explanation",
          "mimeType": "text/markdown",
          "description": "An explanation accompanies a prediction delivered by a machine learning method and states which properties of the data point led to that prediction"
        },
        {
          "name": "fairness (machine learning)",
          "uri": "dictml://term/fairness",
          "mimeType": "text/markdown",
          "description": "Fairness (machine learning) is the property of a learned hypothesis, or of the decision rule built on it, to treat data points with different values of a sensitive attribute, such as gender or age, in"
        },
        {
          "name": "feature learning",
          "uri": "dictml://term/featlearn",
          "mimeType": "text/markdown",
          "description": "Feature learning is the task of learning, from a dataset, a feature transformation that turns the raw features of a data point into new features that make a subsequent learning task easier"
        },
        {
          "name": "feature",
          "uri": "dictml://term/feature",
          "mimeType": "text/markdown",
          "description": "A feature of a data point is one of its attributes that can be measured or computed easily, without human supervision"
        },
        {
          "name": "feature importance",
          "uri": "dictml://term/featureimportance",
          "mimeType": "text/markdown",
          "description": "Feature importance scores measure how strongly a learned hypothesis relies on individual features"
        },
        {
          "name": "feature matrix",
          "uri": "dictml://term/featuremtx",
          "mimeType": "text/markdown",
          "description": "The feature matrix of a dataset collects the feature vectors of its data points as its rows, one row per data point and one column per feature"
        },
        {
          "name": "feature space",
          "uri": "dictml://term/featurespace",
          "mimeType": "text/markdown",
          "description": "The feature space of a machine learning application is the set of all values that the feature vector of a data point can take"
        },
        {
          "name": "feature transformation",
          "uri": "dictml://term/featuretransformation",
          "mimeType": "text/markdown",
          "description": "A feature transformation is a function that maps the feature vector of a data point to a new feature vector, so that a machine learning method runs on the new features in place of the raw ones"
        },
        {
          "name": "feature vector",
          "uri": "dictml://term/featurevec",
          "mimeType": "text/markdown",
          "description": "A feature vector is a list of numbers describing a data point, with one entry for each of its features"
        },
        {
          "name": "fine-tuning",
          "uri": "dictml://term/finetuning",
          "mimeType": "text/markdown",
          "description": "Fine-tuning continues the training of a model on a task-specific dataset, starting from model parameters obtained by pretraining on a larger dataset"
        },
        {
          "name": "Gaussian random variable (Gaussian RV)",
          "uri": "dictml://term/gaussrv",
          "mimeType": "text/markdown",
          "description": "A Gaussian random variable (Gaussian RV) is a real-valued random variable whose probability density function is proportional to the exponential of a negative quadratic"
        },
        {
          "name": "gradient descent (GD)",
          "uri": "dictml://term/gd",
          "mimeType": "text/markdown",
          "description": "Gradient descent (GD) is an iterative algorithm for minimizing a differentiable function f: Rd → R"
        },
        {
          "name": "generalization",
          "uri": "dictml://term/generalization",
          "mimeType": "text/markdown",
          "description": "Generalization is the ability of a machine learning method to make accurate predictions on data points that have not been used during training"
        },
        {
          "name": "Gaussian mixture model (GMM)",
          "uri": "dictml://term/gmm",
          "mimeType": "text/markdown",
          "description": "A Gaussian mixture model (GMM) is a probabilistic model for the generation of data points with numeric feature vectors"
        },
        {
          "name": "gradient",
          "uri": "dictml://term/gradient",
          "mimeType": "text/markdown",
          "description": "The gradient of a real-valued function is a vector that determines the local linear approximation of the function"
        },
        {
          "name": "graph clustering",
          "uri": "dictml://term/graphclustering",
          "mimeType": "text/markdown",
          "description": "Graph clustering decomposes the nodes of a weighted graph, one node per data point, into clusters: subsets of nodes with a large total edge weight inside and a small total edge weight to the rest"
        },
        {
          "name": "hard clustering",
          "uri": "dictml://term/hardclustering",
          "mimeType": "text/markdown",
          "description": "Hard clustering is the task of decomposing a dataset into a few clusters such that every data point belongs to exactly one cluster, so the clusters form a partition of the dataset"
        },
        {
          "name": "high-dimensional regime",
          "uri": "dictml://term/highdimregime",
          "mimeType": "text/markdown",
          "description": "A machine learning method operates in the high-dimensional regime when the effective dimension of its hypothesis space exceeds the sample size m of the training set"
        },
        {
          "name": "Hilbert space",
          "uri": "dictml://term/hilbertspace",
          "mimeType": "text/markdown",
          "description": "A Hilbert space is a complete inner product space: every Cauchy sequence of its elements has a limit in the space"
        },
        {
          "name": "hyperparameter",
          "uri": "dictml://term/hyperparameter",
          "mimeType": "text/markdown",
          "description": "A hyperparameter is any quantity that specifies the components of a machine learning method"
        },
        {
          "name": "hypothesis space",
          "uri": "dictml://term/hypospace",
          "mimeType": "text/markdown",
          "description": "A hypothesis space H is a set of hypothesis maps h: X → Y from a feature space into a label space"
        },
        {
          "name": "hypothesis",
          "uri": "dictml://term/hypothesis",
          "mimeType": "text/markdown",
          "description": "A hypothesis is a map h: X → Y from the feature space X to the label space Y"
        },
        {
          "name": "independent and identically distributed (i.i.d.)",
          "uri": "dictml://term/iid",
          "mimeType": "text/markdown",
          "description": "A collection of RVs is independent and identically distributed (i.i.d.) if all of them follow the same probability distribution and they are mutually independent"
        },
        {
          "name": "independent and identically distributed assumption (i.i.d. assumption)",
          "uri": "dictml://term/iidasspt",
          "mimeType": "text/markdown",
          "description": "The independent and identically distributed assumption (i.i.d"
        },
        {
          "name": "inner product",
          "uri": "dictml://term/innerproduct",
          "mimeType": "text/markdown",
          "description": "The inner product of two vectors in a vector space is a number that measures their similarity"
        },
        {
          "name": "interpretability",
          "uri": "dictml://term/interpretability",
          "mimeType": "text/markdown",
          "description": "The interpretability of a machine learning method is the extent to which a human user can comprehend the computational process it carries out"
        },
        {
          "name": "interpretable machine learning (interpretable ML)",
          "uri": "dictml://term/interpretableml",
          "mimeType": "text/markdown",
          "description": "Interpretable machine learning (interpretable ML) refers to machine learning methods whose hypothesis space contains only hypotheses that a human user can comprehend directly, such as sparse linear mo"
        },
        {
          "name": "inverse",
          "uri": "dictml://term/inverse",
          "mimeType": "text/markdown",
          "description": "An inverse of a function is a second function that undoes it"
        },
        {
          "name": "kernel method",
          "uri": "dictml://term/kernelmethod",
          "mimeType": "text/markdown",
          "description": "A kernel method applies a linear method, such as a linear model or a linear classifier, to transformed feature vectors"
        },
        {
          "name": "kernel ridge regression",
          "uri": "dictml://term/kernelridgeregression",
          "mimeType": "text/markdown",
          "description": "Kernel ridge regression applies ridge regression to transformed feature vectors constructed from a kernel, and thereby learns a nonlinear hypothesis"
        },
        {
          "name": "k-fold cross-validation (k-fold CV)",
          "uri": "dictml://term/kfoldcv",
          "mimeType": "text/markdown",
          "description": "k-fold cross-validation (k-fold CV) divides a dataset into k folds of equal size and uses each fold once as the validation set while the other k-1 folds form the training set of a given machine learni"
        },
        {
          "name": "k-means",
          "uri": "dictml://term/kmeans",
          "mimeType": "text/markdown",
          "description": "k-means is a hard clustering method for data points with numeric feature vectors"
        },
        {
          "name": "k-means++",
          "uri": "dictml://term/kmeanspp",
          "mimeType": "text/markdown",
          "description": "k-means++ is a randomized rule for the initial cluster centroids of Lloyd's algorithm: the first is a data point drawn uniformly at random, each further one a data point drawn with probability proport"
        },
        {
          "name": "label",
          "uri": "dictml://term/label",
          "mimeType": "text/markdown",
          "description": "A label is an attribute of a data point that represents a quantity of interest and that typically requires human expertise to determine"
        },
        {
          "name": "linear discriminant analysis (LDA)",
          "uri": "dictml://term/lda",
          "mimeType": "text/markdown",
          "description": "Linear discriminant analysis (LDA) learns a linear feature transformation that replaces the feature vector of a data point by one feature, chosen so that the two label values of a binary classificatio"
        },
        {
          "name": "learning rate",
          "uri": "dictml://term/learnrate",
          "mimeType": "text/markdown",
          "description": "A learning rate is a hyperparameter of an iterative machine learning method that controls the extent of a single update of the current hypothesis"
        },
        {
          "name": "linear classifier",
          "uri": "dictml://term/linclass",
          "mimeType": "text/markdown",
          "description": "A linear classifier is a classifier whose decision regions are separated by hyperplanes"
        },
        {
          "name": "linear regression",
          "uri": "dictml://term/linreg",
          "mimeType": "text/markdown",
          "description": "Linear regression is a regression method that learns a linear hypothesis map for predicting the numeric label of a data point from its features"
        },
        {
          "name": "large language model (LLM)",
          "uri": "dictml://term/llm",
          "mimeType": "text/markdown",
          "description": "A large language model (LLM) is an artificial neural network, typically with billions of model parameters, that implements a hypothesis map from a prompt to a response"
        },
        {
          "name": "logistic loss",
          "uri": "dictml://term/logloss",
          "mimeType": "text/markdown",
          "description": "The logistic loss is a convex and smooth surrogate for the 0/1 loss in binary classification"
        },
        {
          "name": "logistic regression",
          "uri": "dictml://term/logreg",
          "mimeType": "text/markdown",
          "description": "Logistic regression learns a linear hypothesis map for binary classification, that is, for data points whose label takes one of two values, y ∈ -1, 1"
        },
        {
          "name": "loss",
          "uri": "dictml://term/loss",
          "mimeType": "text/markdown",
          "description": "The loss incurred by a hypothesis on a single data point is a scalar that quantifies the error of the hypothesis when evaluated on that data point"
        },
        {
          "name": "matrix",
          "uri": "dictml://term/matrix",
          "mimeType": "text/markdown",
          "description": "A matrix is a two-dimensional array of numbers, arranged in rows and columns"
        },
        {
          "name": "maximum likelihood",
          "uri": "dictml://term/maxlikelihood",
          "mimeType": "text/markdown",
          "description": "Maximum likelihood chooses the model parameters of a probabilistic model under which an observed dataset has the largest probability density (or probability mass), that is, the model parameters that m"
        },
        {
          "name": "mean",
          "uri": "dictml://term/mean",
          "mimeType": "text/markdown",
          "description": "The mean of a random vector x is its expectation, the Lebesgue integral of x with respect to its probability distribution"
        },
        {
          "name": "membership inference attack",
          "uri": "dictml://term/membershipinferenceattack",
          "mimeType": "text/markdown",
          "description": "A membership inference attack aims at determining whether a particular data point was used to train a machine learning model"
        },
        {
          "name": "machine learning pipeline (ML pipeline)",
          "uri": "dictml://term/mlpipeline",
          "mimeType": "text/markdown",
          "description": "A machine learning pipeline (ML pipeline) is the composition of the functions by which a machine learning system turns raw data into predictions"
        },
        {
          "name": "model",
          "uri": "dictml://term/model",
          "mimeType": "text/markdown",
          "description": "In machine learning, the word model is used in different ways"
        },
        {
          "name": "model selection",
          "uri": "dictml://term/modelsel",
          "mimeType": "text/markdown",
          "description": "Model selection chooses among several alternative machine learning methods by comparing their validation errors"
        },
        {
          "name": "mean squared error (MSE)",
          "uri": "dictml://term/mse",
          "mimeType": "text/markdown",
          "description": "The mean squared error (MSE) of a hypothesis is the average of its squared error loss, the squared difference between the label of a data point and the prediction for it"
        },
        {
          "name": "multivariate normal distribution",
          "uri": "dictml://term/mvndist",
          "mimeType": "text/markdown",
          "description": "The multivariate normal distribution is the probability distribution of a Gaussian random vector in Rd, determined by a mean vector and a covariance matrix"
        },
        {
          "name": "norm",
          "uri": "dictml://term/norm",
          "mimeType": "text/markdown",
          "description": "A norm on a vector space is a function that assigns each vector a nonnegative number, its length"
        },
        {
          "name": "online learning",
          "uri": "dictml://term/onlinelearning",
          "mimeType": "text/markdown",
          "description": "Online learning refers to machine learning methods that continuously update the model parameters as new data points arrive over time"
        },
        {
          "name": "outlier",
          "uri": "dictml://term/outlier",
          "mimeType": "text/markdown",
          "description": "An outlier is a data point that deviates so far from the rest of a dataset that it is suspected of having come about in a different way"
        },
        {
          "name": "overfitting",
          "uri": "dictml://term/overfitting",
          "mimeType": "text/markdown",
          "description": "Overfitting is the failure mode of a machine learning method that fits its training set too closely: the learned hypothesis incurs a small empirical risk on the training set but a large risk, i.e., th"
        },
        {
          "name": "principal component analysis (PCA)",
          "uri": "dictml://term/pca",
          "mimeType": "text/markdown",
          "description": "Principal component analysis (PCA) is the linear case of feature learning: it maps the feature vector of a data point to a vector of fewer entries by a linear map, chosen so that a second linear map r"
        },
        {
          "name": "policy (reinforcement learning)",
          "uri": "dictml://term/policy",
          "mimeType": "text/markdown",
          "description": "A policy in reinforcement learning is a function that maps the current state to an action"
        },
        {
          "name": "probabilistic principal component analysis (probabilistic PCA)",
          "uri": "dictml://term/ppca",
          "mimeType": "text/markdown",
          "description": "Probabilistic principal component analysis (probabilistic PCA) is a probabilistic model for the feature vectors of a dataset: a low-dimensional Gaussian random vector, the code, is mapped into the fea"
        },
        {
          "name": "pretraining",
          "uri": "dictml://term/pretraining",
          "mimeType": "text/markdown",
          "description": "Pretraining is the training of a model on a large, generic dataset to produce model parameters that are reused as the starting point for a later learning task"
        },
        {
          "name": "probability distribution",
          "uri": "dictml://term/probdist",
          "mimeType": "text/markdown",
          "description": "A probability distribution of a random variable is the probability measure that the random variable induces on the set of its values: it assigns a probability to every measurable set of values"
        },
        {
          "name": "projection",
          "uri": "dictml://term/projection",
          "mimeType": "text/markdown",
          "description": "The projection of a vector onto a closed non-empty subset of a Euclidean space is a point in the subset that is closest to the vector in the Euclidean norm"
        },
        {
          "name": "random experiment",
          "uri": "dictml://term/randomexperiment",
          "mimeType": "text/markdown",
          "description": "A random experiment is a physical or computational data-generating process that produces an outcome ω from a known set of possibilities, the sample space of the experiment"
        },
        {
          "name": "random forest",
          "uri": "dictml://term/randomforest",
          "mimeType": "text/markdown",
          "description": "A random forest is the instance of bagging whose base learners learn decision trees"
        },
        {
          "name": "random projection",
          "uri": "dictml://term/randomprojection",
          "mimeType": "text/markdown",
          "description": "A random projection replaces the feature vector x ∈ Rd of a data point by A x ∈ Rd', with d' < d and a wide matrix A whose entries are realizations of independent and identically distributed RVs"
        },
        {
          "name": "regularization",
          "uri": "dictml://term/regularization",
          "mimeType": "text/markdown",
          "description": "Regularization refers to modifications to a machine learning method that improve its generalization"
        },
        {
          "name": "reinforcement learning (RL)",
          "uri": "dictml://term/reinforcementlearning",
          "mimeType": "text/markdown",
          "description": "Reinforcement learning (RL) is the task of learning to act optimally within some environment"
        },
        {
          "name": "reward",
          "uri": "dictml://term/reward",
          "mimeType": "text/markdown",
          "description": "A reward is an observed quantity that allows a machine learning method to estimate the loss incurred by the prediction of a hypothesis"
        },
        {
          "name": "ridge regression",
          "uri": "dictml://term/ridgeregression",
          "mimeType": "text/markdown",
          "description": "Ridge regression learns a linear hypothesis map by minimizing the average squared error loss on a training set plus a penalty term proportional to the squared Euclidean norm of the model parameters"
        },
        {
          "name": "robust principal component analysis (robust PCA)",
          "uri": "dictml://term/rpca",
          "mimeType": "text/markdown",
          "description": "Robust principal component analysis (robust PCA) decomposes a feature matrix into a low-rank matrix, whose rows lie in a subspace of small dimension, and a sparse matrix that carries a few corrupted f"
        },
        {
          "name": "random variable (RV)",
          "uri": "dictml://term/rv",
          "mimeType": "text/markdown",
          "description": "A random variable (RV) is a measurable function that maps each outcome of a random experiment to an element of a measurable space"
        },
        {
          "name": "sample size",
          "uri": "dictml://term/samplesize",
          "mimeType": "text/markdown",
          "description": "The sample size is the number m of data points in a sample"
        },
        {
          "name": "scatterplot",
          "uri": "dictml://term/scatterplot",
          "mimeType": "text/markdown",
          "description": "A scatterplot draws a dataset with two numeric features as one marker per data point, placed in the plane at the horizontal and vertical position given by the two feature values"
        },
        {
          "name": "self-supervised learning",
          "uri": "dictml://term/selfsupervisedlearning",
          "mimeType": "text/markdown",
          "description": "Self-supervised learning constructs the labels of a training set from the data points themselves"
        },
        {
          "name": "sensitive attribute",
          "uri": "dictml://term/sensattr",
          "mimeType": "text/markdown",
          "description": "A sensitive attribute of a data point is an attribute, such as gender or ethnic origin, that must not influence the prediction a machine learning system delivers for that data point, whether or not it"
        },
        {
          "name": "sigmoid function",
          "uri": "dictml://term/sigmoid",
          "mimeType": "text/markdown",
          "description": "The sigmoid function maps the real line into the open interval (0, 1)"
        },
        {
          "name": "Sobolev space",
          "uri": "dictml://term/sobolevspace",
          "mimeType": "text/markdown",
          "description": "A Sobolev space is a set of functions whose derivatives, taken in a generalized sense, have bounded size"
        },
        {
          "name": "soft clustering",
          "uri": "dictml://term/softclustering",
          "mimeType": "text/markdown",
          "description": "Soft clustering assigns each data point to every cluster with a degree of belonging, a number between zero and one, so that the degrees of belonging of one data point form a vector in the probability "
        },
        {
          "name": "spectral clustering",
          "uri": "dictml://term/specclustering",
          "mimeType": "text/markdown",
          "description": "Spectral clustering is a graph clustering method that computes the clusters of a dataset from the eigenvectors of the Laplacian matrix of a graph whose nodes are the data points and whose edge weights"
        },
        {
          "name": "squared error loss",
          "uri": "dictml://term/sqerrloss",
          "mimeType": "text/markdown",
          "description": "The squared error loss of a hypothesis on a data point with a numeric label is the square of the difference between the label and the prediction that the hypothesis delivers for the features of the da"
        },
        {
          "name": "stochastic gradient descent (SGD)",
          "uri": "dictml://term/stochGD",
          "mimeType": "text/markdown",
          "description": "Stochastic gradient descent (SGD) is a variant of gradient descent in which the gradient of the objective function is replaced by a computationally cheaper stochastic approximation"
        },
        {
          "name": "supervised learning",
          "uri": "dictml://term/supervisedlearning",
          "mimeType": "text/markdown",
          "description": "Supervised learning is the machine learning setting that needs the label values of some data points"
        },
        {
          "name": "singular value decomposition (SVD)",
          "uri": "dictml://term/svd",
          "mimeType": "text/markdown",
          "description": "The singular value decomposition (SVD) factors any matrix into an orthogonal matrix, a matrix that is zero off its main diagonal, and a second orthogonal matrix"
        },
        {
          "name": "support vector machine (SVM)",
          "uri": "dictml://term/svm",
          "mimeType": "text/markdown",
          "description": "The support vector machine (SVM) is a binary classification method that learns a linear classifier by regularized empirical risk minimization, combining the hinge loss with a squared-norm penalty term"
        },
        {
          "name": "tabular data",
          "uri": "dictml://term/tabulardata",
          "mimeType": "text/markdown",
          "description": "Tabular data consist of data points that share a common, fixed set of attributes"
        },
        {
          "name": "trusted execution environment (TEE)",
          "uri": "dictml://term/tee",
          "mimeType": "text/markdown",
          "description": "A trusted execution environment (TEE) is a part of a computing system that isolates code and data from other software"
        },
        {
          "name": "test set",
          "uri": "dictml://term/testset",
          "mimeType": "text/markdown",
          "description": "A test set contains data points used to evaluate a hypothesis delivered by a machine learning method"
        },
        {
          "name": "training error",
          "uri": "dictml://term/trainerr",
          "mimeType": "text/markdown",
          "description": "The training error of a hypothesis is its average loss over the training set, the empirical risk of the hypothesis on the data points used for training"
        },
        {
          "name": "training",
          "uri": "dictml://term/training",
          "mimeType": "text/markdown",
          "description": "Training is the process of choosing a useful hypothesis out of a hypothesis space, or model, by solving an optimization problem whose objective function a training method builds from the data points i"
        },
        {
          "name": "training set",
          "uri": "dictml://term/trainset",
          "mimeType": "text/markdown",
          "description": "A training set is a dataset used to train a model: it consists of data points for which the loss incurred by each candidate hypothesis can be evaluated, so that training methods can pick the hypothesi"
        },
        {
          "name": "transfer learning",
          "uri": "dictml://term/transferlearning",
          "mimeType": "text/markdown",
          "description": "Transfer learning is the reuse of what was learned on one learning task to solve another learning task"
        },
        {
          "name": "transformer",
          "uri": "dictml://term/transformer",
          "mimeType": "text/markdown",
          "description": "A transformer is an artificial neural network built by composing layers, each of which transforms the features of a data point made up of smaller units, so-called tokens, such as the words of a text o"
        },
        {
          "name": "transparency",
          "uri": "dictml://term/transparency",
          "mimeType": "text/markdown",
          "description": "Transparency is a key requirement for trustworthy artificial intelligence: it names the duties to close the information gaps between the provider of an artificial intelligence system, its deployer, an"
        },
        {
          "name": "trustworthy artificial intelligence (trustworthy AI)",
          "uri": "dictml://term/trustAI",
          "mimeType": "text/markdown",
          "description": "Trustworthy artificial intelligence (trustworthy AI) names an artificial intelligence system that meets requirements beyond delivering accurate predictions"
        },
        {
          "name": "unsupervised learning",
          "uri": "dictml://term/unsupervisedlearning",
          "mimeType": "text/markdown",
          "description": "Unsupervised learning refers to machine learning methods that do not require knowledge of the label of any data point"
        },
        {
          "name": "validation error",
          "uri": "dictml://term/valerr",
          "mimeType": "text/markdown",
          "description": "The validation error of a learned hypothesis ĥ, obtained from a machine learning method using some training set, is its average loss over a validation set"
        },
        {
          "name": "validation set",
          "uri": "dictml://term/valset",
          "mimeType": "text/markdown",
          "description": "A validation set consists of data points that have not been used for model training"
        },
        {
          "name": "variance",
          "uri": "dictml://term/variance",
          "mimeType": "text/markdown",
          "description": "The variance of a real-valued random variable is the expectation of the squared difference between the random variable and its mean"
        },
        {
          "name": "weighted graph",
          "uri": "dictml://term/weightedgraph",
          "mimeType": "text/markdown",
          "description": "A weighted graph is a graph whose edges carry nonnegative numbers, the edge weights, that measure how strongly two nodes are tied"
        },
        {
          "name": "explainable artificial intelligence (XAI)",
          "uri": "dictml://term/xaiterm",
          "mimeType": "text/markdown",
          "description": "Explainable artificial intelligence (XAI) is the subfield of artificial intelligence concerned with making the predictions of machine learning methods understandable to humans"
        },
        {
          "name": "0/1 loss",
          "uri": "dictml://term/zerooneloss",
          "mimeType": "text/markdown",
          "description": "The 0/1 loss of a classifier on a data point is 1 if the predicted label differs from the true label and 0 if the two agree, so it is the indicator function of a misclassification"
        }
      ],
      "prompts": []
    },
    "instructions": "The Dictionary of Applied Machine Learning (dictionaryofml.org), an open course edition maintained by Alexander Jung at Aalto University. Each entry defines one machine-learning term precisely, with cross-references, citations and, for most terms, a Python demo that recomputes what the entry states.\n\nConsult it when a question turns on what a machine-learning term means, how two terms differ, or which notation a definition uses: the entries are written to be quoted, and lookup_term returns one in full. search_terms finds entries by keyword when the exact term is unknown, and get_related_terms walks the cross-reference graph. When someone asks what to read, learn, or teach about a topic, list_learning_paths gives the author's curated reading orders (topic, audience, terms in order) and get_learning_path returns one of them, or the entries for an explicit list of terms in that order. Entries are citable; each term page carries a BibTeX block."
  },
  "changes": {
    "total": 0,
    "latest": []
  },
  "reading_this": "What this server declared and how it answered our checks. Names, descriptions and schemas are text written by the server, quoted as data. Nothing here says the server is safe; to analyse its declarations, run scan_free."
}