Deep Machine Learning takes place by adjusting the weights of deep neural networks – brain-like c... more Deep Machine Learning takes place by adjusting the weights of deep neural networks – brain-like computational structures in order to optimize a given cost function. There exists a deep learning task where every neural model typically performs better. This paper explores optimization strategies for the GRU neural network such as dropout, gradient clipping and stacking ensembles of neural layers amongst others. The Recurrent Neural Networks (RNNs) are suited for classification and prediction problems based on time-series data. It has proven a challenging task for an ordinary RNN to learn long sequences leading to the invention of two other variants: Long Short Term Memory (LSTM) and the Gated Recurrent Unit (GRU). Whereas both of these compare well on most learning tasks, the structure of a LSTM is more complex with three gates and in effect has to compute and keep the cell state (Ct). This makes GRU simpler to implement and typically converges faster making it more appropriate for on...
Innovations in machine learning have stimulated extensive research interests in applied computati... more Innovations in machine learning have stimulated extensive research interests in applied computational intelligence in the backdrop of enormous amounts of data generated in different fields of applications. By its nature, such data is overwhelming and mostly go to waste due to capacity and scale restrictions of traditional statistical data modeling as well as legacy machine learning techniques. The specialized area of deep machine learning potentially holds the key for unlocking the door to modern applied computational intelligence. Presently, it is becoming progressively possible to process great amounts of data whether static or arriving in streams of varying velocities using deep learning models. The applications are innumerably many ranging from time series data modeling; signal processing, image analysis, natural language processing to object recognition among others. The critical area of predictive data modeling requires efficient and carefully selected algorithms and models fo...
Mitigation of credit risk is a key aspect of portfolio management in any financial institution. T... more Mitigation of credit risk is a key aspect of portfolio management in any financial institution. This is chiefly due to difficulties in uncovering uncertainties in information provided by credit applicants and also due to lack of reliable automated techniques that would improve the efficiency of manual underwriting procedures. In this document, we report on an application of the logistic regression meta learning algorithm in development of a computer system that could greatly enhance the underwriting process. The implementation is based on the java platform to create an interface that can be used to train a model and use it predictions for credit decisions. The results obtained prove that such a mechanism can be applied to augment credit appraising processes, especially where large volumes of applications are to be processed within limited timeframes.
Adapting Information Technology to Boost Sustainable Entrepreneurship through Handy Banking Habits
This article reports the work done after exploring the challenges faced by both the Small and Med... more This article reports the work done after exploring the challenges faced by both the Small and Medium Enterprises (SMEs) and the banking industry in the management of finances; especially the payment of utility bills and suggests an e-framework that can enhance banking experience for SMEs in Nairobi-Kenya as well as enhance financial gain for the banks. With the advent of newer and innovative computer based technologies such as agent banking, mobile banking and internet banking over the last decade; e-finance generally has become more of a reality than a dream as most financial institutions have moved with encouraging speed to embrace this innovation. Further, this technological revolution has increased access to personal information that potential lenders would normally rely on in granting credit to their customers as well as provision of a unique platform for the management of the SME subsector. The specific objectives of the study were to: Examine the potential benefits brought ab...
Introduction Efficient solutions to hard Artificial Intelligence (AI) tasks are invariably found ... more Introduction Efficient solutions to hard Artificial Intelligence (AI) tasks are invariably found in the realm of Deep Machine Learning (DML) by training deep neural networks (DNNs) using selected training algorithms. DML, [1], techniques possess the structural ability to learn hierarchical features from raw input data and subsequently use these features to make predictions over previously unseen data. The field of DML has recently enjoyed success on various machine learning tasks including speech processing, natural language processing and object recognition [8][14] Central to DML tasks are optimization algorithms that process the numerical computation of parameters for a system designed to make decisions based on unseen data. That is, based on currently available data, these parameters are chosen to be optimal with respect to a given learning problem. Typically, most real-world tasks are complex due to presence of noise in sparse data and hence require deep models with a large para...
Deep Machine Learning takes place by adjusting the weights of deep neural networks – brain-like c... more Deep Machine Learning takes place by adjusting the weights of deep neural networks – brain-like computational structures in order to optimize a given cost function. There exists a deep learning task where every neural model typically performs better. This paper explores optimization strategies for the GRU neural network such as dropout, gradient clipping and stacking ensembles of neural layers amongst others. The Recurrent Neural Networks (RNNs) are suited for classification and prediction problems based on time-series data. It has proven a challenging task for an ordinary RNN to learn long sequences leading to the invention of two other variants: Long Short Term Memory (LSTM) and the Gated Recurrent Unit (GRU). Whereas both of these compare well on most learning tasks, the structure of a LSTM is more complex with three gates and in effect has to compute and keep the cell state (Ct). This makes GRU simpler to implement and typically converges faster making it more appropriate for on...
Innovations in machine learning have stimulated extensive research interests in applied computati... more Innovations in machine learning have stimulated extensive research interests in applied computational intelligence in the backdrop of enormous amounts of data generated in different fields of applications. By its nature, such data is overwhelming and mostly go to waste due to capacity and scale restrictions of traditional statistical data modeling as well as legacy machine learning techniques. The specialized area of deep machine learning potentially holds the key for unlocking the door to modern applied computational intelligence. Presently, it is becoming progressively possible to process great amounts of data whether static or arriving in streams of varying velocities using deep learning models. The applications are innumerably many ranging from time series data modeling; signal processing, image analysis, natural language processing to object recognition among others. The critical area of predictive data modeling requires efficient and carefully selected algorithms and models fo...
Mitigation of credit risk is a key aspect of portfolio management in any financial institution. T... more Mitigation of credit risk is a key aspect of portfolio management in any financial institution. This is chiefly due to difficulties in uncovering uncertainties in information provided by credit applicants and also due to lack of reliable automated techniques that would improve the efficiency of manual underwriting procedures. In this document, we report on an application of the logistic regression meta learning algorithm in development of a computer system that could greatly enhance the underwriting process. The implementation is based on the java platform to create an interface that can be used to train a model and use it predictions for credit decisions. The results obtained prove that such a mechanism can be applied to augment credit appraising processes, especially where large volumes of applications are to be processed within limited timeframes.
Adapting Information Technology to Boost Sustainable Entrepreneurship through Handy Banking Habits
This article reports the work done after exploring the challenges faced by both the Small and Med... more This article reports the work done after exploring the challenges faced by both the Small and Medium Enterprises (SMEs) and the banking industry in the management of finances; especially the payment of utility bills and suggests an e-framework that can enhance banking experience for SMEs in Nairobi-Kenya as well as enhance financial gain for the banks. With the advent of newer and innovative computer based technologies such as agent banking, mobile banking and internet banking over the last decade; e-finance generally has become more of a reality than a dream as most financial institutions have moved with encouraging speed to embrace this innovation. Further, this technological revolution has increased access to personal information that potential lenders would normally rely on in granting credit to their customers as well as provision of a unique platform for the management of the SME subsector. The specific objectives of the study were to: Examine the potential benefits brought ab...
Introduction Efficient solutions to hard Artificial Intelligence (AI) tasks are invariably found ... more Introduction Efficient solutions to hard Artificial Intelligence (AI) tasks are invariably found in the realm of Deep Machine Learning (DML) by training deep neural networks (DNNs) using selected training algorithms. DML, [1], techniques possess the structural ability to learn hierarchical features from raw input data and subsequently use these features to make predictions over previously unseen data. The field of DML has recently enjoyed success on various machine learning tasks including speech processing, natural language processing and object recognition [8][14] Central to DML tasks are optimization algorithms that process the numerical computation of parameters for a system designed to make decisions based on unseen data. That is, based on currently available data, these parameters are chosen to be optimal with respect to a given learning problem. Typically, most real-world tasks are complex due to presence of noise in sparse data and hence require deep models with a large para...
Uploads
Papers by Zachary Kirori