IIT Guwahati Junior Research Fellow Interview Questions, Process, and Tips

Train semi supervised machine learning models by using a combination of labeled and unlabeled data.

• Start by training a model on a small amount of labeled data

• Use the trained model to make predictions on the unlabeled data

• Incorporate the predictions into the training set and retrain the model

• Repeat the process until the model reaches a satisfactory level of performance

Parametric types of machine learning are algorithms that make assumptions about the functional form of the relationship between inputs and outputs.

• Parametric models have a fixed number of parameters that are learned from the training data.

• Examples include linear regression, logistic regression, and linear SVM.

• They are often simpler and faster to train compared to non-parametric models.

• Parametric models are suitable for...read more

Types of Machine Learning include supervised learning, unsupervised learning, semi-supervised learning, reinforcement learning, and self-supervised learning.

• Supervised Learning: The model is trained on labeled data.

• Unsupervised Learning: The model is trained on unlabeled data.

• Semi-Supervised Learning: A combination of labeled and unlabeled data is used for training.

• Reinforcement Learning: The model learns through trial...read more

Eigen value is a scalar associated with a square matrix that represents how a transformation stretches or compresses space along its eigenvectors.

• Eigen values are solutions to the characteristic equation det(A - λI) = 0, where A is the matrix, λ is the eigen value, and I is the identity matrix.

• They represent the factor by which the eigenvector is scaled during the transformation.

• Eigen values can be real or complex numb...read more

Eigen vector is a vector that does not change its direction when a linear transformation is applied to it.

• Eigen vectors are used in linear algebra to understand the behavior of linear transformations.

• They represent directions along which a linear transformation has a simple effect, such as scaling.

• Eigen vectors are associated with eigenvalues, which represent the scaling factor of the eigenvector.

• For example, in a 2x2 ...read more

PCA is a mathematical technique used for dimensionality reduction by finding the principal components of a dataset.

• PCA involves calculating the eigenvectors and eigenvalues of the covariance matrix of the data.

• The eigenvectors represent the directions of maximum variance in the data, while the eigenvalues indicate the amount of variance along each eigenvector.

• The principal components are the eigenvectors corresponding ...read more

For a small dataset of 300 samples, a neural network with one hidden layer would be more suitable for better accuracy.

• A neural network with one hidden layer is simpler and less prone to overfitting on a small dataset.

• With a small dataset, a complex neural network with more hidden layers may lead to overfitting and poor generalization.

• A neural network with one hidden layer can capture the basic patterns in the data effe...read more

Yes, a neural network can accept complex numbers as input.

• Neural networks can be designed to accept complex numbers as input by using complex-valued weights and activations.

• Complex-valued neural networks have been used in applications such as signal processing and image recognition.

• Complex numbers can represent both magnitude and phase information, making them useful for certain types of data.

• Complex-valued neural netw...read more

kmeans algorithm is a clustering algorithm that partitions data into k clusters based on similarity.

• Divides data points into k clusters based on distance from centroid

• Iteratively assigns data points to nearest centroid and updates centroids

• Converges when centroids no longer change significantly

• Commonly used in machine learning for clustering data points

Evaluation metrics are used to measure the performance or effectiveness of a system, project, or process.

• Evaluation metrics can include quantitative measures such as accuracy, precision, recall, F1 score, and AUC-ROC.

• They can also include qualitative measures such as user satisfaction, usability, and user engagement.

• Evaluation metrics help in assessing the success of a project or system and identifying areas for improv...read more

Machine learning is a subset of artificial intelligence that focuses on developing algorithms to make predictions based on data, while deep learning is a subset of machine learning that uses neural networks to learn from large amounts of data.

• Machine learning involves developing algorithms that can learn from and make predictions or decisions based on data.

• Deep learning is a subset of machine learning that uses neural ...read more

Selecting k value in kmeans algorithm involves using techniques like elbow method and silhouette score.

• Use the elbow method to find the point where the rate of decrease sharply shifts, indicating the optimal k value.

• Calculate silhouette score for different k values and choose the one with the highest score.

• Consider domain knowledge and the specific problem requirements when selecting k value.

• Experiment with different k...read more

Precision and recall can be calculated using values from a confusion matrix.

• Precision = TP / (TP + FP)

• Recall = TP / (TP + FN)

• Where TP = True Positive, FP = False Positive, FN = False Negative

• For an n x n confusion matrix, sum the values in each row and column to get TP, FP, and FN for each class