Schedule

Bharat Light & Power Pvt Ltd (BLP Group) is a leading renewable energy generation and technology company in India. My talk will revolve around how AI, ML and Augmented Analytics are being employed by our organization for renewable energy, manufacturing and utilities sector for downtime reduction, preventive maintenance, energy management, bringing about automation, inventory optimization, manufacturing cycle time reduction and several other use cases. At BLP Clean Energy, we employ analytics and ML for a holistic approach for renewable energy asset performance monitoring, analysis and prediction of impending failures of components of assets. Our approach is a confluence of machine learning for prediction of assets’ behavioural trends in future, streaming data mining to spot and notify underperformance as and when it occurs and rule based methods for decision-making. Our predictive intelligence solutions for failure prediction enable prevention of component failures in several cases and in others, readiness at wind and solar farms to handle potential upcoming component failures. This insight in conjunction with estimates of the remaining useful life of components, through machine learning, converts unplanned downtime into planned downtime. Through real time big data analytics on our cognitive computing platform, we ensure remote, automated, regular health assessment of assets. Our analytics solutions enable domain experts on an ongoing basis to chart out precise action plans to alleviate damage from predicted failures and tackle root causes of underperformance which are communicated to Operations and Management teams at the farms through our closed loop workflow on our AI platform.

In the Digital Era, technology evolves every day and changes the way we live & work. What does it take to remain relevant in such a changing landscape? While it is important for organizations to provide reskilling opportunities to bridge the ever growing talent gap, one’s own ability to learn will go a long way to survive & thrive in the world of Data, Analytics and AI.

AI isn’t dangerous, but the human bias is! However, one must be cognizant about any unintended consequences of using this technology. The unseen harm that AI can cause is them reflecting our human biases in the data sets the organisations collect. There would be a compounding effect during the emergence of AI, as the algorithms exhibit self-learning from data. To fight this prejudice, I would talk about aspects like diversity inclusive mindset for data collection, the possibility of building and taking AI to an elevated level and recognising the vitality involved with AI training systems.

By 2020, every Samsung product will be AI enabled, creating entirely new experiences and value for users. In this talk, we describe how the fundamental technologies such as automatic speech recognition and natural language understanding are being used to create these new experiences. Bixby is Samsung's AI Platform the core of which is Bixby Voice – an intelligent, personalized Assistant that enables users to use Voice for Samsung or 3rd party services. Bixby learns, evolves, and adapts to you to help you get things done seamlessly. Each feature of Bixby is designed to make your life easier, whether it's through touch, type, or voice. We talk about the evolution of Bixby Voice & associated research & development work at Samsung (including R&D centers in India). We talk about challenges in domain & intent classification for Natural Language understanding, and Deep Learning based solutions ranging from CNN vs. RNNs, Word vs. Character based models, Handling of variations of data and conflicts between data. We also describe recent efforts and results in improving speech recognition (ASR) performance for end users of Samsung's Bixby voice assistant. One of the main challenges with ASR is speaker variability. Different users, depending on their style, accent and other traits can pronounce the same word or sentence in many different ways. Hence, a one-size-fits-all approach works poorly for some segment of users, even though aggregate ASR performance may be quite high. Specifically, we present two techniques of improving ASR performance by considering the end user. The first is implemented at a macro level and uses a Hierarchical Accent Determination system dealing with accent variations. For English, eight accents [GB, US, Australian, Canadian, Spanish, Korean, and Indian & Chinese] are identified at macro level and accent-specific models corresponding to the identified accents are used. Similar approaches are applicable for languages like Chinese, Hindi, Arabic, which have significant accent variations at regional level. The second technique involves adapting the ASR models for a particular end user. Here, we describe how the acoustic model can be enhanced for a particular user's speaking style by training a new model on-device for that user. The key challenges in this problem are how to select utterances for training and adaptation, how to generate ground truth labels and handle errors in ground truth estimation.

This session would cover how the enterprises are transforming themselves using machine learning and AI that augments the traditional process of analytics in an organization. Augmented analytics helps business users to take relevant actions from insights that are automatically generated by the system. This session would also cover how SAP portfolio and strategy enables businesses to have an un-biased decision making process and helps them to identify the correct set of business drivers.

AI will pass on the biases of its creators and the data its creators feed it. If we want there to be a woman’s perspective in the new world of AI, we need women to be part of it

Data-driven decision making using Artificial Intelligence is the trend of the data, seeing applications in a variety of domains. In this talk, Geetha will talk about an novel and noble application of AI in the area of healthcare developed at NIRAMAI – a noninvasive, affordable and accessible solution to detect early stage breast cancer.

Explainable AI refers to techniques in artificial intelligence (AI) that explains the decision taken by ‘black box’ machine learning and deep learning models. It is imperative that decisions made by AI models are understood so as to remove biases in decision making and make the models trust-worthy. The explanations can be created either through global interpreters that explain the models at feature importance level or through local interpreters that explain why a particular decision was taken for a specific data point.

The talk will focus on my learnings so far in these 15 years.I will talk about my experiences working across multiple problems and multiple types of data. The challenges, opportunities and how the data science field itself has grown tremendously in a decade.