Claims:We Claim: 1. A processor-implemented method (300) comprising steps of: receiving (302), via an input/output interface, a multivariate mixed-variable time series data of a plurality of sensory observations, cluster-labels associated with the multivariate mixed-variable time series data, and a cluster-independent random noise, wherein multivariate mixed-variable time series data comprises continuous and discrete feature variables; pre-processing (304), via one or more hardware processors, the received multivariate mixed-variable time series data, wherein the pre-processing comprising steps of: normalizing the continuous feature variables by bounding heterogeneous measurements between a predefined range using a min-max scaling technique; and transforming the discreate feature variables by representing as a sparse binary vector using a one-hot encoding technique. training (306), via the one or more hardware processors, a plurality of neural networks of a privacy preserving generative adversarial network (ppGAN) in a first phase and a second phase using pre-processed multivariate mixed-variable time series data, wherein the plurality of neural networks including an embedding neural network, a recovery neural network (120), a generator neural network (122), a critic neural network (124) and a discriminator neural network (126); providing (308), via the one or more hardware processors, a test data to generate a synthetic private dataset for data disclosure using the trained plurality of neural networks of the ppGAN; and estimating (310), via the one or more hardware processors, an identifiability of the multivariate mixed-variable time series data from the generated synthetic private dataset, wherein the estimation satisfies a predefined process-identifiability criteria. 2. The processor-implemented method of claim 1, wherein the first phase training of the plurality of neural networks of the ppGAN comprising steps of: training (402), via the one or more hardware processors, the embedding neural network using a predefined low-dimensional mixed feature training dataset to obtain a high-dimensional mixed feature embeddings; training (404), via the one or more hardware processors, the recovery neural network (120) using the obtained high-dimensional mixed feature embeddings to reconstruct the low-dimensional mixed feature dataset; training (406), via the one or more hardware processors, the supervisor neural network using the obtained high-dimensional mixed-feature embeddings for a single step ahead predictions of the high-dimensional mixed-feature embeddings, wherein the supervisor neural network is utilized to model a temporal dynamics of the low-dimensional mixed feature training dataset; and training (408), via the one or more hardware processors, the critic neural network (124) using the high-dimensional mixed feature embeddings to predict a target high-dimensional feature embedding, wherein the critic neural network (124) is utilized to model the relationship between independent and dependent variables of the low-dimensional mixed feature training dataset. 3. The processor-implemented method of claim 1, wherein a second phase training of the plurality of neural networks of a ppGAN comprising steps of: transforming (502), via the one or more hardware processors, cluster-independent random noise using one or more cluster-labels associated with a predefined training dataset to obtain a cluster-dependent random noise; performing (504), via the one or more hardware processors, a linear transformation on a concatenation of the low-dimensional mixed feature training dataset and the cluster-dependent random noise to obtain a synthetic-private noise; training (506), via the one or more hardware processors, the generator neural network (122) using the obtained synthetic-private noise to obtain a high-dimensional synthetic-private mixed feature embeddings; training (508), via the one or more hardware processors, the critic neural network (124) using the high-dimensional synthetic-private mixed feature embeddings to predict the synthetic-private target feature embedding; training (510), via the one or more hardware processors, the discriminator neural network (126) using the high-dimensional synthetic-private mixed feature embeddings to assign a label, wherein the discriminator neural network (126) classifies the high-dimensional synthetic-private mixed feature embeddings as fake; training (512), via the one or more hardware processors, the supervisory neural network (128) using the high-dimensional synthetic-private mixed feature embeddings to generate a single-step ahead predictions of the high-dimensional synthetic-private mixed feature embeddings; and training (514), via the one or more hardware processors, the recovery neural network (120) using the single-step ahead high-dimensional synthetic-private mixed feature embeddings to obtain the low-dimensional synthetic-private mixed feature dataset. 4. The processor-implemented method of claim 1, wherein the low-dimensional mixed feature validation dataset is utilized for the hyper-parameter tuning of the ppGAN. 5. A system (100) comprising: an input/output interface (104) to a multivariate mixed-variable time series data of a plurality of sensory observations, cluster-labels associated with the multivariate mixed-variable time series data, and a cluster-independent random noise, wherein multivariate mixed-variable time series data comprises continuous and discrete feature variables; a memory (110) in communication with the one or more hardware processors (108), wherein the one or more hardware processors are configured to execute programmed instructions stored in the memory to: pre-process the received multivariate mixed-variable time series data, wherein the pre-process includes normalizing the continuous feature variables by bounding heterogeneous measurements between a predefined range through a min-max scaling technique, and transforming discreate feature variables by representing as a sparse binary vector through a one-hot encoding technique; train a plurality of neural networks of a privacy preserving generative adversarial network (ppGAN) in a first phase and a second phase using pre-processed multivariate mixed-variable time series data, wherein the plurality of neural networks including an embedding neural network, a recovery neural network (120), a generator neural network (122), a critic neural network (124) and a discriminator neural network (126); provide a test data to generate a synthetic private dataset for data disclosure using the trained plurality of neural networks of the ppGAN; and estimate an identifiability of the multivariate mixed-variable time series data from the generated synthetic private dataset, wherein the estimation satisfies a predefined process-identifiability criteria. 6. The system (100) of claim 5, wherein the low-dimensional mixed feature validation dataset is utilized for the hyper-parameter tuning of the ppGAN. 7. A non-transitory computer readable medium storing one or more instructions which when executed by one or more processors on a system, cause the one or more processors to perform method comprising: receiving, via an input/output interface, a multivariate mixed-variable time series data of a plurality of sensory observations, cluster-labels associated with the multivariate mixed-variable time series data, and a cluster-independent random noise, wherein multivariate mixed-variable time series data comprises continuous and discrete feature variables; pre-processing, via one or more hardware processors, the received multivariate mixed-variable time series data, wherein the pre-processing comprising steps of: normalizing the continuous feature variables by bounding heterogeneous measurements between a predefined range through a min-max scaling technique; and transforming the discreate feature variables by representing as a sparse binary vector using a one-hot encoding technique. training, via the one or more hardware processors, a plurality of neural networks of a privacy preserving generative adversarial network (ppGAN) in a first phase and a second phase using pre-processed multivariate mixed-variable time series data , wherein the plurality of neural networks including an embedding neural network, a recovery neural network (120), a generator neural network (122), a critic neural network (124) and a discriminator neural network (126); providing, via the one or more hardware processors, a test data to generate a synthetic private dataset for data disclosure using the trained plurality of neural networks of the ppGAN; and estimating, via the one or more hardware processors, an identifiability of the multivariate mixed-variable time series data from the generated synthetic private dataset, wherein the estimation satisfies a predefined process-identifiability criteria. Dated this 4th Day of March 2022 Tata Consultancy Services Limited By their Agent & Attorney (Adheesh Nargolkar) of Khaitan & Co Reg No IN-PA-1086 , Description:FORM 2 THE PATENTS ACT, 1970 (39 of 1970) & THE PATENT RULES, 2003 COMPLETE SPECIFICATION (See Section 10 and Rule 13) Title of invention: PRIVACY PRESERVING GENERATIVE MECHANISM FOR INDUSTRIAL TIME-SERIES DATA DISCLOSURE 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 The disclosure herein generally relates to the field of data privacy and more specifically, to a method and system for privacy preserving generative mechanism for data-disclosure of industrial multivariate mixed-variable time series data. BACKGROUND In the era of rapid advances in Artificial Intelligence (AI) deployment of deep learning techniques in the cloud for production, it has several complexities and risks involved relating to privacy, security, fairness, and accountability of data. Usually, regulators and policymakers across the globe present governance protocols such as the General Data Protection Regulation (GDPR), US Health Insurance Portability and Accountability Act (HIPAA), California Consumer Privacy Law (CCPA), European Commission AI Act, etc., to protect the ownership, and confidentiality of sensitive individual user information. These regulations present a Catch-22 of privacy and are mandatory for tech companies to comply with to avoid lawsuits and penalties. These regulations protect the privacy of individuals, encourage anonymization of the sensitive personal information for data-disclosure to be shared with third-parties. The gold standards for security techniques in deep learning include cryptography techniques such as Homomorphic Encryption (HE), Secure Multi-party Computation (SMC), Differential Privacy (DP) & Information-Theoretic Privacy for data disclosure, Federated ML, Ethereum blockchain, and Smart contracts. There’s a growing awareness and interest across several industrial data behemoths such as FMCG, oil & gas, aviation, power, semiconductor engineering, manufacturing etc. to prevent membership inference, model inversion, attribute inference, hyperparameter and parameter inference, and property inference by a third-party (adversary) to access unauthorized process plant operational data, which embeds the trade secrets, the product formulations & simultaneously adopting privacy embedded-AI techniques for digital twins to leverage the big data for process control, optimization, uncertainty quantification, etc. There is a need and necessity for a mathematical framework to enhance privacy-preserving, trade-off to preserve utility for data monetization of the large-scale industrial and manufacturing plants multivariate mixed-variable time series data. The existing techniques of enabling privacy-preserving mechanisms for data disclosure have lack luster utility and suffer from inherent drawbacks of preserving the original data characteristics in the private dataset generated for data-disclosure. SUMMARY Embodiments of the disclosure present technological improvements as solutions to one or more of the above-mentioned technical problems recognized by the inventors in conventional systems. For example, in one embodiment, a method and system for privacy preserving generative mechanism for data-disclosure of industrial data is provided. In one aspect, a processor-implemented method for privacy preserving generative mechanism for data-disclosure of industrial data is provided. The method includes one or more steps such as receiving, via an input/output interface, a multivariate mixed-variable time series data of a plurality of sensory observations, the cluster-labels associated with the multivariate mixed-variable time series data and a cluster-independent random noise, pre-processing, via one or more hardware processors, the received multivariate mixed-variable time series data, training, via a one or more hardware processors, a plurality of neural networks of a privacy preserving adversarial neural network architecture in two phases, providing, via the one or more hardware processors, a test data to generate a synthetic private dataset for data disclosure using the trained privacy preserving adversarial neural network architecture, and estimating, via the one or more hardware processors, identifiability of the multivariate mixed-variable time series data from the generated synthetic private dataset. In another aspect, a system for privacy preserving generative mechanism for data-disclosure of industrial data is provided. The system includes an input/output interface configured to receive a multivariate mixed-variable time series data of a plurality of sensory observations, the cluster-labels associated with the multivariate mixed-variable time series data and a cluster-independent random noise, one or more hardware processors and at least one memory storing a plurality of instructions, wherein the one or more hardware processors are configured to execute the plurality of instructions stored in the at least one memory. Further, the system is configured to pre-process the received a multivariate mixed-variable time series data, wherein the pre-process includes normalizing continuous feature variables by bounding heterogeneous measurements between a predefined range through a min-max scaling technique; and transforming discreate feature variables by representing as a sparse binary vector through a one-hot encoding technique. Further, the system is configured to train a plurality of neural networks of a privacy preserving adversarial neural network architecture in two phases, provide a test data to generate a synthetic private dataset for data disclosure using the trained privacy preserving adversarial neural network architecture and estimate an identifiability of the multivariate mixed-variable time series data from the generated synthetic private dataset, wherein the estimation satisfies a predefined process-identifiability criteria. In yet another aspect, one or more non-transitory machine-readable information storage mediums are provided comprising one or more instructions, which when executed by one or more hardware processors causes a method for privacy preserving generative mechanism for data-disclosure of industrial data is provided. The method includes one or more steps such as receiving, via an input/output interface, a multivariate mixed-variable time series data of a plurality of sensory observations, the cluster-labels associated with the multivariate mixed-variable time series data and a cluster-independent random noise, pre-processing, via one or more hardware processors, the received multivariate mixed-variable time series data, training, via a one or more hardware processors, a plurality of neural networks of a privacy preserving adversarial neural network architecture in two phases, providing, via the one or more hardware processors, a test data to generate a synthetic private dataset for data disclosure using the trained privacy preserving adversarial neural network architecture, and estimating, via the one or more hardware processors, identifiability of the multivariate mixed-variable time series data from the generated synthetic private dataset. It is to be understood that the foregoing general descriptions and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed. BRIEF DESCRIPTION OF THE DRAWINGS 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: FIG. 1 illustrates a block diagram of an exemplary system for privacy preserving generative mechanism for data-disclosure of industrial data, according to an embodiment of the present disclosure. FIG. 2 is a functional block diagram of the system for privacy preserving generative mechanism for data-disclosure of industrial data, according to an embodiment of the present disclosure. FIG. 3 is a flow diagram to illustrate a method for privacy preserving generative mechanism for data-disclosure of industrial data, in accordance with some embodiments of the present disclosure. FIG. 4 is a flow diagram to illustrate a first phase training of a plurality of neural networks of a privacy preserving adversarial neural network architecture, in accordance with some embodiments of the present disclosure. FIG. 5 is a flow diagram to illustrate a second phase training of the plurality of neural networks of the privacy preserving adversarial neural network architecture, in accordance with some embodiments of the present disclosure. DETAILED DESCRIPTION OF EMBODIMENTS 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 scope of the disclosed embodiments. The embodiments herein provide a method and system for privacy preserving generative mechanism for data-disclosure of the industrial multivariate mixed-variable time-series data. The proposed invention fuses an industrial multivariate mixed-variable time series data with a random gaussian noise to preserve the privacy of the industrial multivariate mixed-variable time series data and trades-off the privacy with the utility of synthetic-private data. It is to be noted that in the era of rapid advances in Artificial Intelligence (AI) techniques, deployment of deep learning techniques in the cloud for production has several complexities and risks involved relating to privacy, security, fairness, and accountability. There is a need and necessity to develop privacy embedded physics-informed deep learning-based generative mechanism to solve complex science and engineering industrial problems by leveraging multivariate synthetic-private data and with a mutual gain of avoiding privacy-breach from adversaries to prevent inference attacks based on de-anonymization techniques. Notable works in the multivariate mixed-variable time series, be short of incorporating privacy-preserving mechanisms into generative models to maintain the end-user trust. The key challenges are determining the exact translation between the regulatory documents and implementable data-driven anonymization of AI techniques. Therefore, the one or more embodiments herein presents the privacy-preserving synthetic industrial data generative mechanism for data disclosure with minimal risk of AI technique and strong privacy guarantees. Embedding privacy by design into the generative mechanism approaches present an alternate paradigm of learning in contrast to the reduced-order modeling and numerical solutions of the industrial data based on the principles in continuum mechanics for data disclosure with privacy. The deep learning-based generative techniques integrated synergistically with privacy-preserving mechanisms to generate the synthetic-private data almost but not-quite real data. The privacy enabled generative techniques provides trustworthiness on the generated privacy-preserving synthetic data that quench both the ethical and legal needs for securing industrial data confidentiality (minimize the process identifiability) and trade-off to preserve the temporal dynamics, probability distributions of the real data to retain utility in downstream predictive analytics tasks. Privacy preservation by sanitizing the data of the associated industrial client (data owner) is of paramount importance for privacy-first tech companies (trusted party holding the sensitive data) to avoid legal subpoena for privacy-breach and it permits the deep-learning practitioners to build AI techniques by utilizing synthetic-private data to improve the modeling of engineering systems. A privacy-preserving Generative Adversarial Network (ppGAN) is a private learning technique that minimizes and regulates the information flow. It is based on a rubric defined on the optimal transport for machine learning in a Kantorovich formulation between the empirical probability distributions of a synthetic-private data and a non-linear original data. The ppGAN comprises a generator neural network that sanitizes the multivariate mixed-variable time series dataset to a certain extent. It occludes threat of inference attacks on the process-plants database, and a discriminator neural network that attempts to maximize the Wasserstein distance between the real data and privacy-preserving synthetic data probability distributions. Referring now to the drawings, and more particularly to FIG. 1 through 3, 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. FIG. 1 illustrates a block diagram of a system (100) for privacy preserving generative mechanism for data-disclosure of the industrial data, in accordance with an example embodiment. Although the present disclosure is explained considering that the system (100) is implemented on a server, it may be understood that the system (100) may comprise one or more computing devices (102), such as a laptop computer, a desktop computer, a notebook, a workstation, a cloud-based computing environment and the like. It will be understood that the system (100) may be accessed through one or more input/output interfaces 104-1, 104-2... 104-N, collectively referred to as I/O interface (104). Examples of the I/O interface (104) may include, but are not limited to, a user interface, a portable computer, a personal digital assistant, a handheld device, a smartphone, a tablet computer, a workstation, and the like. The I/O interface (104) are communicatively coupled to the system (100) through a network (106). In an embodiment, the network (106) may be a wireless or a wired network, or a combination thereof. In an example, the network (106) can be implemented as a computer network, as one of the different types of networks, such as virtual private network (VPN), intranet, local area network (LAN), wide area network (WAN), the internet, and such. The network (106) may either be a dedicated network or a shared network, which represents an association of the different types of networks that use a variety of protocols, for example, Hypertext Transfer Protocol (HTTP), Transmission Control Protocol/Internet Protocol (TCP/IP), and Wireless Application Protocol (WAP), to communicate with each other. Further, the network (106) may include a variety of network devices, including routers, bridges, servers, computing devices, storage devices. The network devices within the network (106) may interact with the system (100) through communication links. The system (100) supports various connectivity options such as BLUETOOTH®, USB, ZigBee, and other cellular services. The network environment enables connection of various components of the system (100) using any communication link including Internet, WAN, MAN, and so on. In an exemplary embodiment, the system (100) is implemented to operate as a stand-alone device. In another embodiment, the system (100) may be implemented to work as a loosely coupled device to a smart computing environment. Further, the system (100) comprises at least one memory with a plurality of instructions, one or more databases (112), and one or more hardware processors (108) which are communicatively coupled with the at least one memory to execute a plurality of modules (114) therein. The components and functionalities of the system (100) are described further in detail. Referring FIG. 2, illustrates a block diagram (200) of the system (100) for privacy preserving generative mechanism for data-disclosure of the industrial data. Herein, the one or more I/O interfaces (104) are configured to receive a multivariate mixed-variable time series data of a plurality of sensory observations, the cluster-labels associated with the multivariate mixed-variable time series data and a cluster-independent random noise. The received multivariate mixed-variable time series data is pre-processed to normalize continuous feature variables by bounding heterogeneous measurements between a predefined range of a min-max scaling technique. Discreate feature variables are transformed by representing as a sparse binary vector through a one-hot encoding technique. In another embodiment, the system (100) comprises a plurality of neural networks (116) of the privacy preserving generative adversarial network architecture. Herein, the plurality of neural networks includes an embedding neural network (118), a recovery neural network (120), a generator neural network (122), a critic neural network (124), a discriminator neural network (126), and supervisory neural network (128). The plurality of neural networks (116) are trained in two phases. In the first phase the embedding neural network (118) is trained using a predefined low-dimensional mixed feature training dataset to obtain a high-dimensional mixed feature embeddings. Wherein, the obtained high-dimensional mixed feature embeddings are used to train a supervisory neural network (128) for a single step ahead predictions of the high-dimensional mixed-feature embeddings. It is to be noted that the supervisory neural network (128) is utilized to model a temporal dynamics of the low-dimensional mixed feature training dataset. Furthermore, the critic neural network (124) of the privacy preserving generative adversarial network architecture is trained using the high-dimensional mixed feature embeddings to predict a target high-dimensional feature embedding. The embedding neural network (118), E_ppGAN learns a high-dimensional representation, H_(train_(n,1:T_n ) ),?n?{(1,2,…,N)} by transforming the corresponding low-dimensional real sequences, I ~_(train_(n,1:T_n ) ),?n?{(1,2,…,N)}. The embedding neural network (118) assists in effective learning by incorporating the semantics of the mixed-feature variables in its feature embeddings, H_(train_(n,1:T_n ) ). The mathematical description of the embedding module is as follows: E_ppGAN:I_(train_(n,1:T_n ) )??_t¦???_(j=1)^f¦??D_j?H_(train_(n,1:T_n ) )??_t¦??_(j=1)^f¦?H_j,?n???????{(1,2,…,N)} (1) wherein, D_j, H_j denotes the j-th feature variable space & latent embedding vector space, respectively. The recovery neural network (120), R_ppGAN obtains the low-dimensional feature representations, I ~_(train_(n,1:T_n ) ),?n?{(1,2,…,N)} from its corresponding high-dimensional temporal latent variables, H_(train_(n,1:T_n ) ),?n?{(1,2,…,N)}, R_ppGAN:H_(n,1:T_n)^*??_t¦??_(j=1)^f¦??H_j?I_(n,1:T_n)^*??_t¦??_(j=1)^f¦D_j ????,?n???{(1,2,…,N)} (2) wherein, the superscript, ^* denotes for real variables, H_(train_(n,1:T_n ) ), I ~_(train_(n,1:T_n ) ) or for synthetic variables, H ^_(n,1:T_n)^' and I ~_(n,1:T_n ) respectively. e ?_rnn and r ?_rnn are autoregressive neural-net models. They are realized with a unidirectional recurrent neural network with extended memory. e_f, r_f^c , and r_f^d are parameterized by feed-forward neural networks. S,S_m denote the sigmoid & softmax activation function, respectively. The trainable parameters of the embedding neural network (E_ppGAN) and recovery neural network (120) (R_ppGAN) are updated by optimizing a cost function through the joint training of the networks in a supervised-learning approach. It incorporates appropriate inductive bias of reconstruction of the original input, I ~_(train_(n,1:T_n ) ) through by minimizing a supervised loss as described below, L_R=?_(n=1)^N¦??I_(train_(n,1:T_n ) )-I ~_(train_(n,1:T_n ) ) ?_2 ?? (3) In joint training of the generator, supervisor, and the recovery neural networks in unsupervised learning approach in a zero-sum game, the system (100) is configured to minimize the first moment, |?I_1-?I_2 | and second-order moment, |v(s ^_1^2 )-v(s ^_2^2 )| differences, defined between the original data, I_(train_(n,1:T_n ) ) and the synthetic-private data, I ~_(n,1:T_n ) respectively. The lower the difference the synthetic-private data sampled from P ~(I ~_(n,1:T_n)^((1:c,c+1:d))) are more likely to come from the same original data distribution, P(I_(n,1:T_n)^((1:c,c+1:d))). The sample means for real data, I_(train_(n,1:T_n ) ) and synthetic-private data, I ~_(n,1:T_n ) are computed by, ?I_1=1/N ?_(j=1)^f¦??_(n=1)^N¦I_(train_(n,1:T_n ))^((j)) ????I^((f)) and ?I_2=1/N ?_(j=1)^f¦?_(n=1)^N¦??I ~_(n,1:T_n)^((j))?I^((f)) ??. The sample variances, s ^_1^2,s ^_2^2?I^((f)) are evaluated by, s ^_1^2=1/N ?_(j=1)^f¦??_(n=1)^N¦?(I_(train_(n,1:T_n ))^((j))-?I_1^((j)) )^2 ???? (4) and s ^_2^2=1/N ?_(j=1)^f¦?_(n=1)^N¦??(I ~_(n,1:T_n)^((j))-?I_2^((j)) )^2 ??. (5) L_US=|?I_1-?I_2 |+|v(s ^_1^2 )-v(s ^_2^2 )| (6) The ppGAN neural architecture in the unsupervised learning approach trains the joint-transformation network of the private-data generator (G_ppGAN), supervisor network (S_ppGAN), and the recovery neural network (120) (R_ppGAN) to transform samples, Z_(n,1:T_n )~P(Z) into R_ppGAN (S_ppGAN (G_ppGAN (Z_(n,1:T_n ))))~P ~(I ~_(n,1:T_n)^((1:c,c+1:d))) such that P ~(I ~_(n,1:T_n)^((1:c,c+1:d)))˜P(I_(n,1:T_n)^((1:c,c+1:d))). By minimizing, L_US, P(D ^_(n,1:T_n )) learns the underlying probability distributions of the input temporal data, P(D ~_(train_(n,1:T_n ) )). In yet another embodiment, the system (100) is configured for a second phase training of the plurality of neural networks of a privacy preserving adversarial neural network architecture. Wherein, the received cluster-independent random noise is transformed using one or more cluster-labels associated with a predefined training dataset to obtain a cluster-dependent random noise. Further, a linear transformation on a concatenation of the low-dimensional mixed feature training dataset and the cluster-dependent random noise is performed to obtain a synthetic-private noise to train the generator neural network (122) of the privacy preserving adversarial neural network architecture. The generator neural network (122) provides a high-dimensional synthetic-private mixed feature embeddings to train the discriminator neural network (126). It would be appreciated that the discriminator neural network (126) assigns a label to the high-dimensional synthetic-private mixed feature embeddings such as real or fake. In yet another embodiment, the supervisory neural network (128) is trained using the high-dimensional synthetic-private feature embeddings to generate the single-step ahead predictions of the high-dimensional synthetic-private feature embeddings and the recovery neural network (120) is trained using a single-step ahead high-dimensional synthetic-private feature embeddings to obtain the low-dimensional synthetic-private mixed feature dataset. In another embodiment, the supervisor neural network (128), S_ppGAN is leveraged to model the non-linear temporal dynamics of the original data, I_(n,1:T_n ). By integrating the S_ppGAN neural-network in the workflow to generate the synthetic-private data, I ~_(n,1:T_n ). The synthetic-private data captures the complex non-linear temporal dependencies of the original data. The G_ppGAN neural-network of the ppGAN framework generates the synthetic high-dimensional latent embeddings, H ^_(n,1:T_n ). The auto-regressive, S_ppGAN:H_(n,1:t-1)^*?H_(n,t)^*,?n?{1,2,…,N},t?1:T_n takes as input H_(n,1:t-1)^* and predicts the single-step ahead temporal latent embeddings, H_(n,t)^* conditioned on the past latent sequences. It can be presented as, S_ppGAN:H_(n,1:T_n)^*??_t¦?_(j=i)^f¦?H_j?H_(n,1:T_n)^('^* ) ?????_t¦?_(j=i)^f¦H_j ??,?n?{(1,2,…,N)} (7) The ppGAN framework effectively captures the temporal dynamics of the real data by minimizing the supervised loss, L_S=[?_(n=1)^N¦???_t¦??H ~_(train_(n,t) )-S_ppGAN (H_(n,1:t-1)^*)?_2]??? (8) The G_ppGAN by operating in the training-loop perceives the ground-truth high-dimensional latent representations, H ~_(train_(n,1:T_n ) ) from the embedding neural network (118). The supervisor neural network (128) S_ppGAN minimizes the loss, L_S by forcing the H ^_(n,1:T_n ) unflagged by the inaccurate adversary (D_ppGAN) to capture the single-step temporal dynamics of the H ~_(train_(n,1:T_n ) ). In yet another embodiment. the critic neural network (124), F_ppGAN:H_(n,1:T_n)^*?R, is a neural-network function leveraged for predictive modeling which maps the independent mixed-feature variables to the target variable, trained in the supervised learning approach. Here, ? (?^*) refers to H_train^((1:f-1)) or H ^^((1:f-1)). The critic neural network (124) takes as input the realizations of H_(train_(n,1:T_n ))^((1:f-1)) or H ^_(n,1:T_n)^((1:f-1)), ?D^(T_n×(1:f-1)) and outputs, H_(train_(n,1:T_n ))^((T)) or H ^_(n,1:T_n)^((T)), ?D^(T_n ) ,?n?{(1,2,…,N)}. The variable subset selection includes the features attributes from the set, {1,…,f-1}?f in ?H_(n,1:T_n)^({(1,…,f-1)})?^* as input feature variables to the predictive modeler. The last feature variable in ?H_(n,1:T_n)^({(T)})?^* denoted by the superscript, T?f denotes the target variable to predict. The loss function for the target variable prediction is described below, L_F (H_(train_(n,1:T_n ) ),H ^_(n,1:T_n ))=?_(n=1)^N¦(F_ppGAN (H_(train_(n,1:T_n ))^((1:f-1)))-F_ppGAN (H ^_(n,1:T_n)^((1:f-1))))^2 ? (9) Given the independent and identically distributed pairs, (H_(train_(n,1:T_n ))^((1:f-1)),H_(train_(n,1:T_n ))^((T))), the critic neural network (124) learns the original relationship between independent feature variables and the target variable in the real dataset. G_ppGAN generates the relationship preserving synthetic-private data, D ^_(n,1:T_n ) by minimizing the L_F during the adversarial joint training by inducing appropriate inductive learning bias and it generalizes well to the domain. c ?_rnn and s ?_rnn are autoregressive neural-net models and it is implemented with a unidirectional recurrent neural network with extended memory. s_f and c_f are implemented by a feed-forward neural network. Let us assume, Z_(n,1:T_n )?R^(T_n×f),?n?{1,2,…,N} denote the realizations of an f-dimensional Gaussian random variable of finite-length, T_n for a sequence, n with values in the range [0,1) sampled from a gaussian distribution, Z~N(0,1). We embed the cluster labels, C_(train_(n,1:T_n ) )?R^(T_n ),?n?{1,2,…,N} in gaussian noise, Z_(n,1:T_n ) to determine the cluster-dependent random noise, Z_(n,1:T_n)^K. The ground-truth, cluster-membership is computed by an iterative Euclidean distance-based classification technique in data mining to split the unlabeled dataset, I_(train_(n,1:T_n ) ) into K-fixed-apriori disjoint clusters by minimizing the within-cluster sum-of-squares criterion. The system (100) is configured to determine the label embedding vectors, e^c?R^(d^' ), ?c?{1,…,K} from the trainable semantics-aware cluster membership embedding matrix, W?R^(K×d^' ) based on the cluster labels, C ~_(train_(n,1:T_n ) ). d^' is the dimension of the embedding matrix, W. We determine the label matrix, L_(n,1:T_n)^c by concatenating the cluster membership embedding vectors, e^c corresponding to the cluster labels, C_(train_(n,1:T_n ) ). Further, the system (100) is configured to perform the matrix-matrix product of Z_(n,1:T_n ) and the transpose of the label matrix, L_(n,1:T_n)^c to obtain the cluster label aware random noise, Z_(n,1:T_n)^K. The discrete-grouping cluster labels, C_(train_(n,1:T_n ) ) corresponding to the mixed-feature dataset, I_(train_(n,1:T_n ) ) are obtained through the unsupervised learning technique. It is determined by the k-means clustering technique as follows: 1. Randomly initialize cluster centroids, µ_1,µ_2,…,µ_K?D^((f)). 2. C ~_(train_(n,1:T_n ) )={};?n?{1,2,…,N} 3. Repeat until convergence so as to minimize the within-cluster sum of pairwise squared deviations: {For every n, while n=N,?n?{1,2,…,N}, For each n, Assign each observation, I_(train_(n,t) ), t?1:T_n,,?n?{1,2,…,N} the closest cluster-label, m?{1,2,…,K} C_(train_(n,1:T_n ) ):=arg (min)-m ??I_(train_(n,1:T_n ) )-µ_m ??^2 (10) For each m?{1,2,…,K}, refine the cluster-centroids, µ_m µ_m:=(?_(t=1)^(T_n)¦?1{C_(train_(n,1:T_n ) )=m} D ~_(train_(n,t) ) ??)/(?_(t=1)^(T_n)¦1{C ~_(train_(n,1:T_n ) )=m} ?) (11) The generator neural network (122) function takes as input the realizations of the low-dimensional, I_(train_(n,1:T_n ) ), Z_(n,1:T_n)^K and outputs a high-dimensional latent variable, H ^_(n,1:T_n ) G_ppGAN:I_(train_(n,1:T_n ) )×Z_(n,1:T_n)^K?H ^_(n,1:T_n ),?n?{1,2,…,N} (12) The synthetic-private data generative can also be viewed as: G_ppGAN:D^((T_n,f))×[0,1]^((T_n,f))?H^((T_n,f)),?n?{1,2,…,N} (13) The temporal latent embeddings, H ^_(n,1:T_n )?H are computed as: H ^_(n,1:T_n )=G_ppGAN (W_(?^' ) [I_(train_(n,1:T_n ) )?Z_(n,1:T_n)^K]),?n?{1,2,…,N} (14) wherein, W_(?^' ) denotes the learnable parameter. The weight matrix is shared across the sequences, n, ?n?{1,2,…,N}. ? denotes the concatenation operator. g ?_rnn is an autoregressive neural-network model and it is implemented with a unidirectional recurrent neural network with extended memory. g_f is parameterized by a feed-forward neural network. The discriminator neural network (126), D_ppGAN in ppGAN architecture is a classifier to differentiate the real, H_(train_(n,1:T_n ) ) and the synthetic-private data, H ^_(n,1:T_n ) and to minimize the Wasserstein distance between joint distributions, P(I_(n,1:T_n)^((1:c,c+1:d))), P ~(I ~_(n,1:T_n)^((1:c,c+1:d))) on a given metric space, H. The D_ppGAN is defined below, D_ppGAN:H_(n,1:T_n)^*?p_(n,1:T_n,m)^*,p_(n,1:T_n)^*,P(H_(n,1:T_n)^*) (15) The discriminator neural network (126) takes as input the realizations of H_(n,1:T_n)^* and outputs the predicted probability of cluster labels, p_(n,1:T_n,m)^*, the predicted probability of adversarial ground-truth, i.e true/fake, p_(n,1:T_n)^*, the estimated multivariate mixed-feature joint probability distributions, P(?H_(n,1:T_n)^((1:c,c+1:d))?^*) as described below, p_(n,1:T_n,m)^*,p_(n,1:T_n)^*,P(?H_(n,1:T_n)^((1:c,c+1:d))?^*)=D_ppGAN (H_(n,1:T_n)^*),?n?{1,…,N} (16) Wherein, the superscript, * corresponds to H_(train_(n,1:T_n ) ), real latent embeddings, p_(train_(n,1:T_n,m) ),p_(train_(n,1:T_n ) ),P(H_(train_(n,1:T_n ))^((1:c,c+1:d))) or H ^_(n,1:T_n ), synthetic latent embeddings p ^_(n,1:T_n,m),p ^_(n,1:T_n ),P ^(H ^_(n,1:T_n)^((1:c,c+1:d))). The G_ppGAN of the ppGAN framework produces synthetic-private latent embeddings, H ^_(n,1:T_n ) of domain H, by operating on the random noise, Z_(n,1:T_n)^K and the real data, I_(train_(n,1:T_n ) ) and the discriminator neural network (126), D_ppGAN by operating on the adversarial learning latent space, H, tries to differentiate latent temporal embeddings, H_(train_(n,1:T_n ) ) & H ^_(n,1:T_n ) and minimizes the Wasserstein distance between joint distributions. The binary cross-entropy loss for classification of the latent embeddings as true or fake is described by, L_U=1/N ?_(n=1)^N¦[?-(y_(n,1:T_n ) log(p_(train_(n,1:T_n ) ))+(1-y_(n,1:T_n ))log(1-p_(train_(n,1:T_n ) )))+ (y_(n,1:T_n ) log(p ^_(n,1:T_n ))+(1-y_(n,1:T_n ))log(1-p ^_(n,1:T_n )))] (17) wherein, y_(n,1:T_n )?{0,1}^(T_n ),?n?{1,…,N} is the adversarial ground-truth, true or fake data. p_(train_(n,1:T_n ) ),p ^_(n,1:T_n )?[0,1]^(T_n ),?n?{1,…,N} is the predicted probability of the true data, and 1-p_(train_(n,1:T_n ) ) & 1-p ^_(n,1:T_n ),?n?{1,…,N} is the predicted probability of fake data. D_ppGAN tries to minimize, L_U. The G_ppGAN tries to maximize, L_U which helps to learn P ^(I_(n,1:T_n)^((1:c,c+1:d))) that best approximates P(I_(train_(n,1:T_n ))^((1:c,c+1:d))). The prognostics of finite cluster-membership, C_(n,1:T_n)^*?K is a multinomial single-output classification task. The ^*?R^(T_n ) refers to true or fake. For the multinomial classification task of cluster-label prediction, the system (100) is configured to compute an individual loss for each cluster-membership, {1,…,K} per latent sequence, n?{1,…,N} and at each time point t(?1:T_n). Further, the system performs a summation operation of the output over the cluster membership & latent sequences. L_LP=1/N[?_(m=1)^K¦??_(n=1)^N¦?-y_(n,1:T_n)^c log(p_(train_(n,1:T_n,m) ) ) ???+?_(m=1)^K¦???_(n=1)^N¦??y_(n,1:T_n)^c log(p ^_(n,1:T_n,m) )]?? (18) wherein, K denotes the fixed-apriori cluster labels. y_(n,1:T_n)^c is the ground-truth binary value of 1 or 0 if cluster label, m is the true label or false label at time point t(?1:T_n) corresponding to latent sequence, n?{1,…,N}. p_(train_(n,1:T_n,m) ) is the predicted probability for real observation at a time point, t of a data sequence, n?{1,…,N} belongs to cluster, m. p ^_(n,1:T_n,m) is the predicted probability for synthetic latent embedding at time point t of sequence, n?{1,…,N} belonging to cluster membership, m. The cluster membership is determined by, C_(train_(n,1:T_n ))^p:=arg (max)-m [S_m (p_(train_(n,1:T_n,m) ))],m?{1,…,K} (19) C ^_(n,1:T_n ):=arg (max)-m [S_m (p ^_(n,1:T_n,m))],m?{1,…,K} (20) wherein, S_m is the softmax activation function. The D_ppGAN tries to minimize, L_LP whereas G_ppGAN tries to maximize, L_LP. C_(train_(n,1:T_n ))^p denote the predicted cluster membership for real data, D_(train_(n,1:T_n ) ) by D_ppGAN in comparison with the ground-truth, C_(train_(n,1:T_n ) ) determined by the clustering technique. C ~_(n,1:T_n ) denote the predicted cluster-membership labels for the synthetic-private data, I ~_(n,1:T_n ). We also minimize the Wasserstein distance, the cost of the optimal transport plan between the estimates of two probability distributions P(H_(train_(n,1:T_n ))^((1:c,c+1:d))) and P ^(H ^_(n,1:T_n)^((1:c,c+1:d))). The system (100) evaluates the Wasserstein loss, L_W and it is described by, L_W=W(P(H_(train_(n,1:T_n ))^((1:c,c+1:d))),P ^(H ^_(n,1:T_n)^((1:c,c+1:d)))) =?inf?_(?~?(P,P ^)) E_((H_(train_(n,1:T_n ) ),H ^_(n,1:T_n ))~?) [?P(H_(train_(n,1:T_n ))^((1:c,c+1:d)))-P ^(H ^_(n,1:T_n)^((1:c,c+1:d))?] (21) wherein, ?~?(P(H_(train_(n,1:T_n ))^((1:c,c+1:d))),P ^(H ^_(n,1:T_n)^((1:c,c+1:d)))) is the set of all possible joint probability distributions between P(H_(train_(n,1:T_n ))^((1:c,c+1:d))) and P ^(H ^_(n,1:T_n)^((1:c,c+1:d))). The D_ppGAN aims to maximize, L_W whereas G_ppGAN attempts to minimize, L_W. g ?_rnn and d ?_rnn are autoregressive neural-net models. They are implemented with a unidirectional recurrent neural network with extended memory. g_f, d_f,d_f^c are implemented by a feed-forward neural network. The conditional generative neural network architecture, G_ppGAN generates synthetic-private data, I ~_(n,1:T_n ) which must be sufficiently anonymized and non-identical enough in order to minimize the reidentification of the observed data, I_(n,1:T_n ). The finite series of data, I ~_(n,t)?D^((f)) and I ~_(n,t+1)?D^((f)) observed at time points, t and t+1(?{1,…,T_n}) of the same sequence, n or belong to distinct sequences, n?{1,…,N} are non-identicalenough as they are actually measured at different time orders. The minimum(shortest) weighted euclidean distance computed between the observations in real data is the measure for non-identicalenough between the synthetic-private data and observed data. In another embodiment, the system (100) is configured to estimate identifiability of the finite observed data by the finite synthetic-private data. Here, the system defines the ?-process identifiability as, there are lower than ?-fraction observations from the real dataset, I_(train_(n,1:T_n ) ) in the synthetic-private dataset, I ~_(n,1:T_n ) that are mostcertainlynot,non-identicalenough in comparison with the real data, I_(train_(n,1:T_n ) ). A randomized synthetic-private data generator mechanism, G_ppGAN provides the ?-process identifiability, for some ???R^+, for any pair of nearest random variables I_(train_(n,t) ),I_(train_(n,t+1) )?D^((f)), for all subsets S of the function-image of the randomized generative technique, I ~_(n,t)?D^((f)) or I ~_(n,1:T_n ) is ?-process identifiable from I_(train_(n,1:T_n ) ), if the following criteria are satisfied, I_F (I_(n,t),I ~_(n,t))=1/N[I(r ~_(n,t)