While architecting a data-based solution, one needs to approach the problem differently depending on the specific strategy being adopted. In traditional machine learning, the focus is mostly on feature engineering. In DL, the emphasis is shifting to tagging larger volumes of data with less focus on feature development. Similarly, synthetic data is a lot more useful in DL than ML. So, the data strategies can be significantly different. Both approaches require very similar approaches to the analysis of errors. But, in most development processes, those approaches are not followed leading to substantial delay in production times. Hyper parameter tuning for performance improvement requires different strategies between ML and DL solutions due to the longer training times of DL systems. Transfer learning is a very important aspect to evaluate in building any state of the art system whether ML or DL. The last but not the least is understanding the biases that the system is learning. Deeply non-linear models require special attention in this aspect as they can learn highly undesirable features.

In our presentation, we will focus on all the above aspects with suitable examples and provide a framework for practitioners for building ML/DL applications.

Recently, we have heard success stories on how deep learning technologies are revolutionizing many industries. Deep Learning has proven huge success in some of the problems in unstructured data domains like image recognition; speech recognitions and natural language processing. However, there are limited gain has been shown in traditional structured data domains like BFSI. This talk would cover American Express’ exciting journey to explore deep learning technique to generate next set of data innovations by deriving intelligence from the data within its global, integrated network. Learn how using credit card spend data has helped improve credit and fraud decisions elevate the payment experience of millions of Card Members across the globe.

Every organization is now focused on the business or customer data and trying hard to get actionable insights out of it. Most of them are either hiring data scientists or up-skilling their existing developers. However, they do understand the domain or business, relevant data and the impact, but, not essentially excellent in data science programming or cognitive computing. To bridge this gap, IBM brings Watson Machine Learning (WML), which is a service for creating, deploying, scoring and managing machine learning models. WML’s machine learning model creation, deployment, and management capabilities are key components of cognitive applications. The essential feature is the “self-learning” capabilities, personalized and customized for specific persona - may it be the executive or business leader, project manager, financial expert or sales advisor. WML makes the need of cognitive prediction easy with model flow capabilities, where machine learning and prediction can be applied easily with just a few clicks, and to work seamlessly without bunch of coding - for different personas to mark boundaries between developers, data scientists or business analysts. In this session, WML's capabilities would be demonstrated by taking a specific case study to solve real world business problem, along with challenges faced. To align with the developers' community, the architecture of this smart platform would be highlighted to help aspiring developers be aware of the design of a large-scale product.

Modern telecommunication networks are complex, consist of several components, generate massive amounts of data in the form of logs (volume, velocity, variety), and are designed for high reliability as there is a customer expectation of always on network access. It can be difficult to detect network failures with typical KPIs as the problems may be subtle with mild symptoms (small degradation in performance). In this workshop on network anomaly detection we will present the application of multivariate unsupervised learning techniques for anomaly detection, and root cause analysis using finite state machines. Once anomalies are detected, the message patterns in the logs of the anomaly data are compared to those of the normal data to determine where the problems are occurring. Additionally, the error codes in the anomaly data are analyzed to better understand the underlying problems. The data preprocessing methodology and feature selection methods will also be presented to determine the minimum set of features that can provide information on the network state. The algorithms are developed and tested with data from a 4G network. The impact of applying such methods is the proactive detection and root cause analysis of network anomalies thereby improving network reliability and availability.

With limited transparency, governments tend to become less accessible to the public. While data science remains as a dominating market in almost all day-to-day life industries, its possibilities in administration and governance are yet to be exploited. In this presentation, I address how emerging concepts such as open data and big data can be used to strengthen democracies and help governments serve the public better. We will explore the various possible ways big data and open data can be used to bridge income inequalities and implement proper resource and service allocation. We will also be looking at different initiative taken by individuals and communities and see the impact those initiatives have had on aiding governance. We will also emphasize the concept of open governance and government open data.

We are experiencing a tremendous explosion in big data. A significant share of this data is unfit for direct analysis or machine learning. This presentation emphasizes on web scraping with powerful R packages such as httr and tools like XPath.This session will also introduce the principles of data cleaning. By the end of the session, you will be able to import raw data from most websites and transform them into proper robust datasets. In the due course of this session, we would build a robust dataset by implementing the above concepts ready for analysis

Data science refers to the science of extracting useful information from data. Knowledge discovery in data bases, data mining, Information extraction also closely match with data science. Supervised learning,Semi supervised learning,Un supervised learning methodologies are out of Academia and penetrated deep into the industry leading to actionable insights, dashboard driven development, data driven reasoning and so on. Data science has been the buzzword for last few years in industry with only a handful of data scientists around the world. The industry needs more and more data scientists in future to solve problems using statistical techniques. The exponential availability of unstructured data from the web has thrown huge challenges to data scientists to exploit them before driving conclusions.

Now that's overload of information and buzzwords. It all has to start somewhere? Where and how to start? How to get hands dirty rather than just reading books and blogs. Is it really science or just code?. Let's get into code to talk data science.

In this workshop i will show the tools required to do real data science rather than just reading by building real models using Deep neural networks and show live demo of the same. Also share some of the key data science techniques every aspiring data scientist should have to thrive in the industry.

From our Tinder matches to movies we watch on Netflix, we tend to encounter recommendation engines on a day to day basis and with the data explosion in place, the number of recommendation engines at play would increase dramatically. In this talk, we look into the underlying principles of recommendation engines. You will learn about the main types of recommendation engine approaches. By the end of this session, you will have ideas on how each of this approaches can be implemented. You will also be able to understand the pros and cons of both these approaches.