4

The answer from @kiner_shah in the comments has solved it: They have eliminated it because in comparing probability for best outcome, it would just introduce additional division and the divisor p(x) is same for all candidates. For example, if I tell you to compare the largest of the following numbers: 3/2, 5/2, 1/2, you will without thinking pick 5/2 ...


2

And that's all, we can infer P(x|y=c) and P(c) from the data. I don't see where the MLE shows its role. Maximum likelihood estimate is used for this very purpose, i.e. to estimate the conditional probability $p(x_j \mid y)$ and marginal probability $p(y)$ . In Naive Bayes Algorithm ,using the properties of conditional probability, we can estimate the joint ...


2

It would be hard to tell if you don't provide what kind of data/problem you are working on, but LDA works well when data that are grouped in gaussian blobs surrounding centroids while vanilla SVM works well when the data is almost linearly separable and naive bayes works well when your features are relatively independent of each other.


1

Bernoulli naïve Bayes $P(x|c_k) = \prod^{n}_{i=1} p^{x_i}_{ki} (1-p_{ki})^{(1-x_i)}$ Let's examine the example of document classification. Let K different text classes and n different terms that our vocabulary contains. $x_i$ are boolean variables (0, 1) expressing if the $i^{th}$ term exists in document x. x is a vector of dimension n. $P(x|c_k)$ ...


1

In machine learning, we can use all the datasets as training data in a model. But if there are too many data sets, or too much data, and we do not split them up, our model may be not produce acceptable results. Why? Because if the model studies too much training data, it may be overfitted. (Just like when you cram for a test, and get overloaded with ...


1

Your assumption about the test data is not correct completely. Maybe you use the test data to tune your learning algorithm to work better on the test data, but it's not the whole thing. Sometimes you need to know that the ML method is working or not and have a sense about how much does it work! You have other scenarios that you want to evaluate your method: ...


1

I refer you to Prof. Tom Mitchell's (the lecturer in the video) draft chapter as the best explanation I could find. I will try to explain it in layman's terms here. Given a boolean problem, our logistic regression classifier will assign the label $Y=0$ iff $w_0 + \sum_{i=1}^{n}w_{i}X_{i} > 0$ (from the video you posted; eqn. 18 in the draft chapter). ...


1

A perceptron is a linear threshold function. That means it has a weight vector $w$, and it outputs $w \cdot x > t$, where $x$ is the input vector and $t$ the threshold. Naïve Bayes makes the assumption that all features are independent (hence the term naïve). It predicts the most likely class by using Bayesian probability, for each class multiplying the ...


1

Naive Bayes is a generative algorithm while Perceptron is a discriminative algorithm. That is the main difference.


1

Your approach would definitely work. I would recommend training a variety of classifiers and comparing their performance using multiclass roc analysis. Also, think about other useful features in addition to the ones you mentioned (e.g. pos tag). Feature engineering is one of the most important factors in building good predictive models. Another thing to keep ...


Only top voted, non community-wiki answers of a minimum length are eligible