FORM 2 THE PATENTS ACT, 1970 (39 of 1970) & THE PATENT RULES, 2003 COMPLETE SPECIFICATION (See Section 10 and Rule 13) Title of invention: RECURRENT NEURAL NETWORK ARCHITECTURE BASED CLASSIFICATION OF ATRIAL FIBRILLATION USING SINGLE LEAD ECG Applicant Tata Consultancy Services Limited A company Incorporated in India under the Companies Act, 1956 Having address: Nirmal Building, 9th floor, Nariman point, Mumbai 400021, Maharashtra, India Preamble to the description The following specification particularly describes the invention and the manner in which it is to be performed. TECHNICAL FIELD [001] The disclosure herein generally relates to identifying abnormal heart rhythm, and, more particularly, to systems and processor implemented methods for recurrent neural network architecture based classification of Atrial Fibrillation (AF) using single lead electrocardiogram (ECG). BACKGROUND [002] Atrial Fibrillation (AF) is an electrophysiological disorder caused when abnormal electrical impulses suddenly start firing in the atria. The heart’s normal rhythm goes awry, resulting in an abnormally fast heart rate with an enhanced risk of stroke and heart attack. Being one of the most common type of arrhythmias, AF is associated with significant mortality and morbidity. Presence of AF affects the electrocardiogram (ECG) morphology and is detected by cardiologists via visual inspection. However, manual detection of intermittent AF episodes from long duration ECG recordings is challenging and often impractical. [003] State of the art techniques of detection of AF from ECG use signal processing and machine learning techniques. Available techniques broadly belong to two categories, 1) atrial analysis based approaches and 2) ventricular response based approaches. Atrial analysis based approaches look for absence of P waves or the presence of fibrillatory f-waves in ECG, whereas ventricular response based approaches analyze irregularities in heart rate over a period of time from R-R interval distances for classification of AF. Atrial activity based approaches are known to be more accurate but they are vulnerable to background noise. Features derived from scatter plot of successive R-R intervals using Poincare and Lorenz plots have been used in some prior art for detection of AF. Again, some prior art relies on statistical features extracted from R-R intervals to measure irregularity in heart rate caused due to AF. However, these techniques are designed for analyzing long ECG recordings. [004] There have been recent developments wherein classification of normal, AF and other abnormal rhythms using short (less than one minute) single lead ECG recordings have been attempted using a combination of classical and deep learning approaches. However, when single lead ECG recordings of short duration are used, accuracy of classification is a challenge. SUMMARY [005] Embodiments of the present disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. [006] In an aspect, there is provided a processor implemented method for classification of atrial fibrillation (AF) comprising the steps of: acquiring, by one or more hardware processors, a single lead electrocardiogram (ECG) recorded for a first pre-defined time period being less than a minute; obtaining, by the one or more hardware processors, a first time series being an R-R interval time series based on R peaks in the acquired ECG; identifying, by the one or more hardware processors, a region having a second pre-defined time period before each of the R peaks to form a second time series, wherein the second time series is a region corresponding to P wave time series and the second pre-defined time period is in the range of 120-200 milliseconds; inputting, by the one or more hardware processors, the first time series and the second time series independently to an associated Long short-term memory (LSTM) network; merging, by the one or more hardware processors, output states of the LSTM network associated with the first time series and the second time series along with a pre-defined set of handcrafted statistical features computed from the acquired ECG to create a composite feature set for classification of the AF; and classifying the AF in the acquired ECG based on the composite feature set using a classifier. [007] In another aspect, there is provided a system for classification of atrial fibrillation (AF) comprising: one or more hardware processors; one or more data storage devices operatively coupled to the one or more hardware processors and configured to store instructions configured for execution by the one or more hardware processors to: acquire a single lead electrocardiogram (ECG) recorded for a first pre-defined time period being less than a minute; obtain a first time series being an R-R interval time series based on R peaks in the acquired ECG; identify a region having a second pre-defined time period before each of the R peaks to form a second time series, wherein the second time series is a region corresponding to P wave time series and the second pre-defined time period is in the range of 120-200 milliseconds; input the first time series and the second time series independently to a pair of Long short-term memory (LSTM) networks; and merge output states of the LSTM network associated with the first time series and the second time series along with a pre-defined set of handcrafted statistical features computed from the acquired ECG to create a composite feature set for classification of the AF; the pair of the LSTM networks, wherein a Bidirectional LSTM (BiLSTM) network and an LSTM network constitute the pair, and wherein the BiLSTM is configured to receive the first time series and perform a temporal analyses of the R-R intervals to capture irregular R-R intervals and the LSTM network is configured to receive the second time series and perform a temporal analyses of atrial activities, wherein the atrial activities include absence of P waves or presence of f-waves before a QRS complex in the acquired ECG; and a classifier comprising a plurality of full connected layers and a softmax function, wherein the classifier is configured to classify the AF in the acquired ECG based on the composite feature set. [008] In yet another aspect, there is provided a computer program product comprising a non-transitory computer readable medium having a computer readable program embodied therein, wherein the computer readable program, when executed on a computing device, causes the computing device to: acquire a single lead electrocardiogram (ECG) recorded for a first pre-defined time period being less than a minute; obtain a first time series being an R-R interval time series based on R peaks in the acquired ECG; identify a region having a second pre-defined time period before each of the R peaks to form a second time series, wherein the second time series is a region corresponding to P wave time series and the second pre-defined time period is in the range of 120-200 milliseconds; input the first time series and the second time series independently to an associated Long short-term memory (LSTM) network comprised therein; merge output states of the LSTM network associated with the first time series and the second time series along with a pre-defined set of handcrafted statistical features computed from the acquired ECG to create a composite feature set for classification of the AF; and classify the AF in the acquired ECG based on the composite feature set using a classifier comprised therein. [009] In accordance with an embodiment of the present disclosure, the first pre-defined time period is 33 seconds. [010] In accordance with an embodiment of the present disclosure, the second pre-defined time period is 200 milliseconds. [011] In accordance with an embodiment of the present disclosure, the region having the second pre-defined time period represents a window before a QRS complex in the acquired ECG where the P wave is located and the second time series comprises a plurality of windows on time axis. [012] In accordance with an embodiment of the present disclosure, the cardinality of the pre-defined set of handcrafted statistical features is 20 [013] In accordance with an embodiment of the present disclosure, the one or more hardware processors are configured to perform a temporal analyses of the R-R intervals using a Bidirectional LSTM (BiLSTM) network to capture irregular R-R intervals; and perform a temporal analyses of atrial activities using an LSTM network, wherein the atrial activities include absence of P waves or presence of f-waves before a QRS complex in the acquired ECG. [014] It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed. BRIEF DESCRIPTION OF THE DRAWINGS [015] The accompanying drawings, which are incorporated in and constitute a part of this disclosure, illustrate exemplary embodiments and, together with the description, serve to explain the disclosed principles. [016] FIG.1 illustrates an exemplary block diagram of a system for recurrent neural network architecture based classification of atrial fibrillation using single lead electrocardiogram (ECG), in accordance with an embodiment of the present disclosure. [017] FIG.2 illustrates an exemplary flow diagram of a computer implemented method for recurrent neural network architecture based classification of atrial fibrillation using single lead ECG, in accordance with an embodiment of the present disclosure. [018] FIG.3A and FIG.3B illustrate an exemplary non-Atrial Fibrillation (AF) and AF ECG waveform respectively, as known in the art. [019] FIG.4 illustrates a network architecture for classifying AF, in accordance with an embodiment of the present disclosure. [020] FIG.5 illustrates a histogram of data length in the PhysioNet 2017 dataset available in the public domain. [021] FIG.6 illustrates cross entropy loss at different epochs on training and validation data, in accordance with an embodiment of the present disclosure. DETAILED DESCRIPTION OF EMBODIMENTS [022] Exemplary embodiments are described with reference to the accompanying drawings. In the figures, the left-most digit(s) of a reference number identifies the figure in which the reference number first appears. Wherever convenient, the same reference numbers are used throughout the drawings to refer to the same or like parts. While examples and features of disclosed principles are described herein, modifications, adaptations, and other implementations are possible without departing from the spirit and scope of the disclosed embodiments. It is intended that the following detailed description be considered as exemplary only, with the true scope and spirit being indicated by the following claims. [023] Systems and methods of the present disclosure address an accurate classification of Atrial Fibrillation (AF) from short single lead electrocardiogram (ECG) recordings. Automatic AF detectors known in the art utilize a limited set of hand-crafted features for designing a classifier. Such AF detectors are designed based on signal processing and machine learning perspective and their outcomes may not be directly clinically interpretable or are vulnerable to noise. Morphological features are extracted from the P-Q, QRS or QT regions of the ECG for detection of AF. Since single lead ECG is used, automatic detection of P-R and P-Q intervals is often difficult, which introduces error in feature computing from those segregated intervals and compromises the performance of the classifier. [024] The approach provided in the present disclosure is driven by clinicians’ views of detecting AF. Mimicking the perception of a clinician is achieved by using a recurrent neural network (RNN) for temporal analyses of R-R intervals and P wave regions and combining them together with a set of statistical features to obtain an improved AF classifier, in accordance with an embodiment of the present disclosure. Particularly, a Long short-term memory (LSTM) network based approach is used that feeds a certain portion of the ECG where the P wave is most likely to be located and allows the classifier to learn the desired features for classification. Hence it is less vulnerable to noise even though single lead ECG recordings are used. [025] Referring now to the drawings, and more particularly to FIG.1 through FIG.6, where similar reference characters denote corresponding features consistently throughout the figures, there are shown preferred embodiments and these embodiments are described in the context of the following exemplary system and/or method. [026] FIG.1 illustrates an exemplary block diagram of a system 100 for recurrent neural network architecture based classification of atrial fibrillation using single lead electrocardiogram (ECG), in accordance with an embodiment of the present disclosure. In an embodiment, the system 100 includes one or more processors 104 being one or more hardware processors or one or more software modules or a combination thereof; communication interface device(s) or input/output (I/O) interface(s) 106; one or more data storage devices or memory 102 operatively coupled to the one or more processors 104; a pair of Long short-term memory (LSTM) networks including a Bidirectional LSTM (BiLSTM) network 108 and an LSTM network 110 described later in the description; and a classifier 112 described later in the description. The one or more processors 104 can be implemented as one or more microprocessors, microcomputers, microcontrollers, digital signal processors, central processing units, state machines, graphics controllers, logic circuitries, and/or any devices that manipulate signals based on operational instructions. Among other capabilities, the processor(s) are configured to fetch and execute computer-readable instructions stored in the memory. In the context of the present disclosure, the expressions ‘processors’ and ‘hardware processors’ may be used interchangeably. In an embodiment, the system 100 can be implemented in a variety of computing systems, such as laptop computers, notebooks, hand-held devices, workstations, mainframe computers, servers, a network cloud and the like. [027] The I/O interface(s) 106 can include a variety of software and hardware interfaces, for example, a web interface, a graphical user interface, and the like and can facilitate multiple communications within a wide variety of networks N/W and protocol types, including wired networks, for example, LAN, cable, etc., and wireless networks, such as WLAN, cellular, or satellite. In an embodiment, the I/O interface(s) can include one or more ports for connecting a number of devices to one another or to another server. [028] The memory 102 may include any computer-readable medium known in the art including, for example, volatile memory, such as static random access memory (SRAM) and dynamic random access memory (DRAM), and/or non-volatile memory, such as read only memory (ROM), erasable programmable ROM, flash memories, hard disks, optical disks, and magnetic tapes. In an embodiment, one or more modules (not shown) of the system 100 can be stored in the memory 102. [029] FIG.2 illustrates an exemplary flow diagram of a computer implemented method 200 for recurrent neural network architecture based classification of atrial fibrillation using single lead ECG, in accordance with an embodiment of the present disclosure. In an embodiment, the system 100 includes one or more data storage devices or memory 102 operatively coupled to the one or more processors 104 and is configured to store instructions configured for execution of steps of the method 200 by the one or more processors 104. The steps of the method 200 will now be explained in detail with reference to the components of the system 100 of FIG.1. Although process steps, method steps, techniques or the like may be described in a sequential order, such processes, methods and techniques may be configured to work in alternate orders. In other words, any sequence or order of steps that may be described does not necessarily indicate a requirement that the steps be performed in that order. The steps of processes described herein may be performed in any order practical. Further, some steps may be performed simultaneously. [030] FIG.3A and FIG.3B illustrate an exemplary non-Atrial Fibrillation (AF) and AF ECG waveform respectively, as known in the art. Clinicians look for two important markers in ECG for AF, viz., (i) absence of P waves or presence of f-waves before the QRS complex, and (ii) irregularly irregular R-R intervals. The method 200 of the present disclosure is designed to capture both these markers based on temporal analyses of the ECG. [031] FIG.4 illustrates a network architecture along with its tensor dimensions at output of different layers for classifying AF, in accordance with an embodiment of the present disclosure. For instance, in the block R-R intervals times series with dimension [None, 1, 66], None indicates no predefined training size, 1 is the dimension of the time series and 66 represents the number of time samples. In sequence models, RNNs use their internal memory to process a sequence of input time series for extraction of temporal patterns that can be used for classification or prediction. LSTM is a class of RNN that can effectively learn a longer pattern of unknown length as it can deal with the exploding and vanishing gradient problems faced by RNNs during training. LSTM does this because of its cell structure that enables deleting less important information from memory. For an input sequence xt = {x1,... ..xT}, an LSTM cell with one forget gate computes a hidden vector sequence ht ={h1,.....hT} by iterating the following equations over t. Here, σ represents the logistic sigmoid function, * denotes element-wise product operation. Input gate, forget gate, output gate and cell activation vectors are represented by i , f , o and c respectively. Hyperbolic tangent is used as the input activation function. [032] In accordance with an embodiment of the present disclosure, the one or more processors 104 are configured to acquire, at step 202, a single lead electrocardiogram (ECG) recorded for a first pre-defined time period being less than a minute. Further at step 204, a first time series is obtained, wherein the first time series is an R-R interval time series based on R peaks in the acquired ECG. In an embodiment, the open source implementation of Behar’s algorithm maybe used for extraction of R peaks to construct the R-R interval time series. In an embodiment, irregular sampling rate of the time series is fixed to 2 Hertz (Hz) using cubic-spline interpolation technique and mapped to the range of 0 to 1. In an embodiment, the first pre-defined time period or the duration of ECG recording is 33 seconds as explained later in the description with reference to the experimental dataset. [033] Since atrial activities behave chaotically during AF, P waves are either not present or there are f-waves in ECG. Although detection of such activities makes an AF classifier more accurate, automatic segmentation of P waves using signal processing approaches is a difficult task due to the varying morphology of P waves. In accordance with the present disclosure, temporal analyses of the atrial activities is performed using the LSTM network 110. The input sequence to the LSTM network 110 is formed by considering a window before the QRS complex in the ECG where the P wave is typically located and stacking multiple such windows on time axis. Accordingly, in an embodiment, the one or more processors 104 are configured, at step 206, to identify a region having a second pre-defined time period before each of the R peaks to form a second time series such that the second time series is a region corresponding to P wave time series. In accordance with the present disclosure, the region having the second pre-defined time period represents a window before a QRS complex in the acquired ECG where the P wave is located and the second time series comprises a plurality of windows on the time axis. Typically, duration of PR interval distance lies between 120-200 milliseconds. Accordingly, in the present disclosure, the second pre-defined time period, is in the range of 120-200 milliseconds. Particularly, in an embodiment, the second pre-defined time period may be 200 milliseconds and it ends 33 milliseconds before a reference R peak. [034] Rapid irregular fluctuation in heart rate is a known symptom of AF. This directly reflects in the ECG, as the R peaks do not repeat after a fixed interval. Clinicians are trained to identify such irregularities via visual inspection. In accordance with the present disclosure, long term temporal dependencies in R-R intervals in an ECG recording are modeled using the BiLSTM network. Both previous and future context of a time series can be effectively utilized in the BiLSTM network, as it processes the input sequence in both forward and backward direction. In accordance with an embodiment of the present disclosure, the one or more processors 104 are configured to input, at step 208, the first time series and the second time series independently to a pair of Long short-term memory (LSTM) networks. In an embodiment, the pair of the LSTM networks include the BiLSTM network 108 and the LSTM network 110. In an embodiment, the BiLSTM network 108 is configured to receive the first time series from step 204 and perform a temporal analyses of the R-R intervals to capture irregular R-R intervals while the LSTM network 110 is configured to receive the second time series from step 206 and perform a temporal analyses of atrial activities, wherein the atrial activities include absence of P waves or presence of f-waves before the QRS complex in the acquired ECG. [035] In accordance with an embodiment of the present disclosure, the one or more processors 104 are configured to merge, at step 210, output states of the LSTM network associated with the first time series and the second time series along with a pre-defined set of handcrafted statistical features computed from the acquired ECG to create a composite feature set for classification of the AF. [036] In an embodiment, 20 handcrafted statistical features are considered to measure the heart rate variability. For instance, entropy related features are used to measure the randomness of a time series. For a time series RRt={RR1,RR2,....RRm}, approximate entropy ApEn(RRt,q,r) is measured in terms of a predefined pattern length (q) and a similarity criterion parameter r . A sequence of vectors{xq(1),xq(2),...xq(m-q+1)in real q -dimensional space is defined from RRt , such that xq(i) ={RRi,RRi+1,RRi+2,... .RRi+q-1}. Two such vectors xq(i) and xq(j) are similar if |RRi+k - RRj+k |