preambleにおける「～において」と「～であって」の違い、米国特許用語集、米国特許翻訳社
トレーニング（training、訓練）とは？
モデルのパラメーターを自動的に調整
学習（learning）とは？
「モデルを作ること。トレーニングデータを入力しながら、適切な出力を行う具体的な計算式／計算方法を求めていく（＝トレーニングしていく）作業」
モデル（model）とは？
「学習後の具体的な計算式／計算方法「学習済みモデル（Learned model、学習モデル）」「トレーニング済みモデル（Trained model、訓練済みモデル）」とも呼ばれる。」
「パラメーターが決まった具体的な式」
「［入力］→［モデル］→［出力］」
「機械学習のモデルを作成（＝学習）する手順／方法は「手法」と呼ばれる。」
「訓練する（学習させる）」
学習：「データから、最もうまく予言できるパラメータを決める」
「データの学習」
「相関関係を見出して、数式化する（モデルを構築する）」
機械学習、Wikipedia
機械学習（きかいがくしゅう、英: machine learning）とは、経験からの学習により自動で改善するコンピューターアルゴリズムもしくはその研究領域で[1][2]、人工知能の一種であるとみなされている。「訓練データ」もしくは「学習データ」と呼ばれるデータを使って学習し、学習結果を使って何らかのタスクをこなす。例えば過去のスパムメールを訓練データとして用いて学習し、スパムフィルタリングというタスクをこなす、といった事が可能となる。
訓練データを使ってプログラムの性能を改善する過程を、「プログラムを訓練する」もしくは「プログラムを学習させる」という。
教師あり学習では訓練データの事を教師データとも呼ぶ
Machine learning, Wikipedia
Machine learning (ML) is a field of inquiry devoted to understanding and building methods that 'learn', that is, methods that leverage data to improve performance on some set of tasks.[1] It is seen as a part of artificial intelligence. Machine learning algorithms build a model based on sample data, known as training data, in order to make predictions or decisions without being explicitly programmed to do so.[2] Machine learning algorithms are used in a wide variety of applications, such as in medicine, email filtering, speech recognition, and computer vision, where it is difficult or unfeasible to develop conventional algorithms to perform the needed tasks.[3]
The discipline of machine learning employs various approaches to teach computers to accomplish tasks where no fully satisfactory algorithm is available. In cases where vast numbers of potential answers exist, one approach is to label some of the correct answers as valid. This can then be used as training data for the computer to improve the algorithm(s) it uses to determine correct answers. For example, to train a system for the task of digital character recognition, the MNIST dataset of handwritten digits has often been used.[10]
Machine learning approaches are traditionally divided into three broad categories, depending on the nature of the "signal" or "feedback" available to the learning system:
Supervised learning: The computer is presented with example inputs and their desired outputs, given by a "teacher", and the goal is to learn a general rule that maps inputs to outputs.
Unsupervised learning: No labels are given to the learning algorithm, leaving it on its own to find structure in its input. Unsupervised learning can be a goal in itself (discovering hidden patterns in data) or a means towards an end (feature learning).
Reinforcement learning: A computer program interacts with a dynamic environment in which it must perform a certain goal (such as driving a vehicle or playing a game against an opponent). As it navigates its problem space, the program is provided feedback that's analogous to rewards, which it tries to maximize.[4]
Supervised learning
Main article: Supervised learning
A support-vector machine is a supervised learning model that divides the data into regions separated by a linear boundary. Here, the linear boundary divides the black circles from the white.
Supervised learning algorithms build a mathematical model of a set of data that contains both the inputs and the desired outputs.[34] The data is known as training data, and consists of a set of training examples. Each training example has one or more inputs and the desired output, also known as a supervisory signal. In the mathematical model, each training example is represented by an array or vector, sometimes called a feature vector, and the training data is represented by a matrix. Through iterative optimization of an objective function, supervised learning algorithms learn a function that can be used to predict the output associated with new inputs.[35] An optimal function will allow the algorithm to correctly determine the output for inputs that were not a part of the training data. An algorithm that improves the accuracy of its outputs or predictions over time is said to have learned to perform that task.[18]
「特許出願における英語翻訳文をより良いものにするために」、米国特許翻訳社
特許英語通信文と英文明細書作成へのガイド（１２）、藤芳寛治、パテント2004、Vol. 57, No. 7
一般英語では名詞を番号で修飾した場合無冠詞であり、特許でもfigure, page, claim等には付けないが、参照番号付きの部材名には冠詞を付ける。
Literal Translation of Patents, Martin Cross
逐語訳の重要性
半導体の素子分離「膜」という表現がピンと来なかったが、面状の膜ではなくて、溝内に残った膜「部分」だと分かった。
「表面を研磨し余分なシリコン酸化膜を除去し、溝の中だけにシリコン酸化膜を残します。」
USJC（ユナイテッド・セミコンダクター・ジャパン株式会社）、FEOL（Front End of Line：基板工程、半導体製造前工程の前半）
1. 素子分離
JP4209206（富士通マイクロエレクトロニクス）
【００２６】
まず、ｐ型シリコン基板１０に、例えばＳＴＩ法により、素子領域を画定する素子分離膜１２を形成する。なお、図３乃至図６は、図２の断面に相当する工程断面図であり、中央の素子分離膜１２により画定された図面右側の素子領域にＮ型トランジスタ形成領域を、図面左側の素子領域にＰ型トランジスタ形成領域を記載している。
in a case in which よりはwhen, whereが良い。
そうは言ってもなんとなくin a case in whichの方が合うような気がする場合もあるような気もしますが。
プラスチックの話 用語集 日精樹脂工業株式会社
見通し線解析：line of sight analysis
見通し解析：＝見通し線解析（？）、sight/visible sight/visibility/view analysis
可視領域解析：viewshed analysis
可視性解析：visibility analysis
ArcGIS Earth facilitates exploration, planning, and analysis in 3D, ArcGIS Earth
New Visibility Analysis Tools in CityEngine 2017.1
Viewshed analysis, Wikipedia
可視性解析、esriジャパン
可視領域解析によるセキュリティの強化、ArcGIS
”ArcGIS における可視性解析は、可視領域解析、見通し解析、スカイライン解析の 3 つに分類することができます。”（「可視性解析とは」）
DBG (dicing before grinding) singulation process
"... During front-end production ... electronic circuits ... are formed on the surface of a silicon wafer ... in back-end production, the wafer backside is thinned and the wafer is singulated by dicing（＊ダイシングにより個片化）. The chips are then encapsulated in a package that will be delivered to end-users." (Silicon wafer thinning, the singulation process, and die strength, DISCO Technical Review Feb. 2016)
US2018012803
"FIG. 1 is a top plan view of four adjacent integrated device cells that have been singulated by a sawing process."
"For example, FIG. 1 is a top plan view of adjacent integrated device cells 13, 13a that have been singulated by a dual cut sawing process with both cuts from the same side. As shown in FIG. 1, for example, a metallic strip 15 having an overall strip width W can be disposed about the periphery（＊周辺、周縁）of the device cells 13, 13a overlapping with the saw streets. A strip width w3 can represent the lateral width of metal remaining on each side of the saw cut after singulation, although of course that width can differ on either side. When the second saw cut passes the exposed metal ends of the metal strip 15, the metal ends can be ripped outwardly from the device cells 13, 13a to form a stringer 35 as shown in FIG. 1. The stringer 35 can be undesirable, as it can fall down onto the circuitry of the device cells 13, 13a and can short out or otherwise damage the device cells 13, 13a. In some embodiments, the stringer 35 can be especially undesirable when the stringer 35 has a length of 90 microns or more, or 100 microns or more. Relatively long stringers 35 (e.g., stringers having a length of 100 microns or more) can increase the chance that the stringer 35 makes contact with the circuitry of the device cells 13, 13a."
EP1178181
"A turbine blade (10) includes an integral airfoil (18), platform (20), shank (22), and dovetail (24), with a pair of holes (36,38) in tandem extending through the platform and shank in series flow communication with an airflow channel (28) inside the shank. Cooling air discharged through the tandem holes effects multiple, convection, impingement, and film cooling using the same air."
当ブログの例文について
本ブログの「特許英語散策」等題した部分では、英語の例文を管理人の独断と偏見で収集し、適宜訳文・訳語を記載しています。
訳文等は原則として対応日本語公報をそのまま写したものです。私個人のコメント部分は（大抵）”＊”を付しています。
訳語は多数の翻訳者の長年の努力の結晶ですが、誤訳、転記ミスもあると思いますのでご注意ください。
