ML Courses (basic level) Flashcards

1
Q

тянуть,

вытащить

A

pull,

pull out

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

польза

A

utility [ютилити]

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

мера качества модели

A

measure of model quality

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

точность модели

A

model accuracy

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

удобство

A

convenience

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

узел,

листовые узлы

A

node,

leaf nodes [лиф ноудс]

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

глубокое дерево,
мелкое дерево

A

deep tree,
shallow tree

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Это проваливает (не в состоянии) охватить все различия в сырых данных

A

It fails to capture as many distinctions in the raw data

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

complicated (усложненный)

A

sophisticated [сэфистикейтэд]

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Это может привести к лучшей производительности

A

It can lead to better performance

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

Рандомный лес обеспечивает гораздо большую предсказательную точность, чем одиночное древо решений

A

The Random Forest provides much better predictive accuracy than a single decision tree

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

ключевые выводы (сущ из глагола take)

A

key takeaways

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

настраивать

A

to tune

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

статистическая техника (как метод)

A

statistical technique [тэкнИк]

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

Признаки, ось X

Метка, ось y

A

features,

label

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

снижаться, расти. Например функция растет, падает

A

to decrease, to increase

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

Уравнение линейной регрессии и что это

A

Linear regression equation helps to find the relationship between x [экс] and y [уай] by fitting a straight [стрэйт] line that best represents the data

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

Коэффициент наклона

A

slope coefficient [коэфишэнт]

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

умноженный на

A

multiplied by

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
20
Q

ускорение, ускорить

A

acceleration [аксэлэрэйшэн],
to accelerate

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
21
Q

быть более устойчивым к выбросам

A

be more robust to outliers

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
22
Q

синоним to tune (настраивать, регулировать)

23
Q

прилагательное: постепенно, с повторениями с повторением итераций

A

iteratively

24
Q

Градиентный спуск и что это

A

Gradient descent is a mathematical technique that iteratively finds the best weights and bias that produce the model with the lowest loss

25
Q

Модель сошлась (типо стала рабочей и тд)

A

The model has converged

26
Q

Сходимость (модели на пример)

A

model convergence [конвёрджэнс]

27
Q

Функции потерь для линейных моделей всегда создают выпуклую поверхность

A

The loss functions for linear models always produce a convex surface

28
Q

свойство

29
Q

Как результат, в результате

A

as a result

30
Q

выпуклость функции потерь

A

convexity of loss function

31
Q

Кривая потерь и что она показывает

A

The loss curve [кёрв] illustrate how loss changes over iterations during training

32
Q

Что такое скорость обучения модели и какова цель при выборе скорости

A

Learning rate is a floating point number, that you set, and that influences how quickly the model converges.

The goal is to pick a learning rate that’s not too high and not too low, so that the model converges quickly

33
Q

Размер выборки для обновления весов. И объяснить что это.

A

Batch Size is a hyperparameter that defines how many examples the model processes before updating its weights and bias.

34
Q

Что такое гиперпараметры и параметры. Key Differences

A

Hyperparameters are values, you choose before training (learning rate, batch size, epochs, etc.).

Parameters are values the model learns during training (weights, bias)

35
Q

Две распространненные техники определения бача, для градиентного спуска. И их объяснение.

A

There are two common gradient descent techniques:

  1. Stochastic [стокхастик] Gradient Descent (SGD)
    - uses only one example per iteration (batch size = 1)
    - Works with enough iterations but introduces noise (fluctuations in loss)
    - Stochastic means that the single example is chosen randomly
  2. Mini-Batch Stochastic Gradient Descent (Mini-Batch SGD)
    - A compromise between full-batch and SGD
    - Uses a small subset(подмножество) of example per batch (1<batch size<dataset size)
    - averages gradients across the batch before updating weights, reducing noise
36
Q

Определение маленького и большого размера батча

A
  1. Small batch size (1-32) behaves like SGD, more noise, but learns faster.
  2. Large batch size (128+) behaves like full-batch gradient descent, more stable but requires more memory.
37
Q

Что такое эпоха?

A

The Epoch [ипок] is a hyperparameter, that is one full pass though the entire training dataset. One epoch is not enough for the model to learn meaningful patterns (значимые закономерности)

38
Q

Что такое логистическая регрессия?

A

Logistic regression is a powerful technique for estimating probabilities. It‘s commonly used in classification problems where the output needs to be a probability between 0 and 1.

39
Q

How does logistic regression work?

A

It takes an input (features) and applies (применять) a sigmoid function to transform it into a probability.

40
Q

What is Sigmoid Function?

A

The sigmoid function ensures (provides) that the model’s output is always between 0 and 1, making it suitable [suitable] (fitting, corresponding) for probability estimation.

41
Q

Обучение модели логистической регрессии, два основных отличение от линейной

A

Training Logistic Regression Models is the same process as training linear regression models, with two key distinctions:

  1. Log [лог] Loss as the loss function.
  2. Regularisation [рэгуларайзэйшн] to prevent overfitting.
42
Q

Почему мы используем log loss для logistic regression?

A

Squared loss works well for a linear model where the rate of change (темп изменения) is constant.

However, the rate of change of a logistic regression model is not constant. The sigmoid curve is s-shaped rather than linear.

43
Q

Точность положительных предсказаний

44
Q

Что такое классификация?

A

Classification is the task of predicting witch of a set of classes (к какому из множества классов) an example belongs to

45
Q

Порог классификации

A

The classification threshold [трэшолд] is used to convert a probability estimate into a definitive class. (окончательный класс)

46
Q

1) если предсказанное значение выше порога

2) если предсказанное значение ниже порога

A

1) If the predicted probability is above the threshold. Then the example is assigned to the positive class.

2) If the predicted probability is below the threshold. Then the example is assigned to the negative class.

47
Q

Более высокий и более низкий порог

A

The choice of classification threshold significantly (существенно) impacts the model’s performance.

A higher threshold increases precision (повышает точность) but may lower recall (меньше полнота класса)

A lower threshold increases recall but may lower precision

48
Q

Для чего используется «матрица путаницы»?

A

A confusion matrix [мэйтрикс] is used to evaluate the performance of a binary classifier

49
Q

Как матрица путаницы делит предсказания

A

A confusion matrix compares the model’s predictions with the actual ground truth, breaking them down into four possible outcomes:
True positive, true negative and false positive, false negative

50
Q

Метрики гипотетической идеальной модели классификации

A

A hypothetical [хайпотэтикал] perfect classification model should have 0 False Positive and 0 False Negative predictions.

If a model achieves this, the performance metrics would be:

Accuracy (точность всех предсказаний) is equal 1

Recall (полнота правильных примеров) is equal 1

False Positive Rate (количество ложных примеров) is equal 0

Precision (точность положительных предсказаний) is equal 1

51
Q

Что такое Receiver-operating characteristic (ROC) curve?

A

ROC [ароси] curve shows how a model’s TPR [типиар] and FPR [эфпиар] change across thresholds (в зависимости от пороговых значений)

52
Q

Что обозначает area under the ROC curve?

A

AUC [эй б си] measures the overall performance of a classifier.

53
Q

От чего зависит оптимальный порог классификации?

A

The optimal threshold depends on the trade-off (компромисс) between false positives and false negatives