3. Problem Statement
To Design a machine learning algorithm capable of guessing
the next number in an integer sequence
1, 2, 3, 4, 5, 7 ?
Guess what's next _???
Example
4. If your answer is 8 (Hint: Power of primes) then you predicted the number correctly.
You used your intuition and tried to find pattern in the integer sequence and then you
developed algorithm which predicted the right number.
That's exactly what we are trying to do the only difference is that the computer will have
to predict the correct number
Problem Statement (Simplified form)
5. Why Sequence Learning ?
• Provide answers to various mathematical and physics problems such as
number theory
• Avoids cases of intense operations such as checking for primeness (used in
cryptography)
• Can be used for predicting next system call(Malware Detection)
6. Sequence Learning Algorithms
These are the algorithms that try to find pattern in the given
sequence
It includes:-
Linear Recurrence Relation
N-gram
RNN(Recurrent Neural Network)
Markov chain
7. Linear Recurrence Relation
A linear recurrence relation is an equation that defines the n’th term in a
sequence in terms of the K previous terms in the sequence.
The recurrence relation is in the form:
8. N gram
An n-gram is a contiguous sequence of n items from a given sequence
of text or speech.
The items can be phonemes, syllables, letters, words or base pairs according
to the application.
The n-grams typically are collected from a text or speech corpus.
When the items are words, n-grams may also be called shingles.