This talk outlines a principle of technical communication which seems simple at first but is devilishly difficult to abide by. It's a principle I try to keep in mind when creating videos aimed at making math and related fields more accessible, and it stands to benefit anyone who regularly needs to describe mathematical ideas in their work. Put simply, it's to resist the temptation to open a topic by describing a general result or definition, and instead let examples precede generality. More than that, it's about finding the type of example which guides the audience to rediscover the general results for themselves. We'll look, aptly enough, at examples of what I mean by this, why it's deceptively difficult to follow, and why this ordering matters.

Since we originally proposed the need for a first-class language, compiler and ecosystem for machine learning (ML) - a view that is increasingly shared by many, there have been plenty of interesting developments in the field. Not only have the tradeoffs in existing systems, such as TensorFlow and PyTorch, not been resolved, but they are clearer than ever now that both frameworks contain distinct "static graph" and "eager execution" interfaces. Meanwhile, the idea of ML models fundamentally being differentiable algorithms – often called differentiable programming – has caught on.

Where current frameworks fall short, several exciting new projects have sprung up that dispense with graphs entirely, to bring differentiable programming to the mainstream. Myia, by the Theano team, differentiates and compiles a subset of Python to high-performance GPU code. Swift for TensorFlow extends Swift so that compatible functions can be compiled to TensorFlow graphs. And finally, the Flux ecosystem is extending Julia’s compiler with a number of ML-focused tools, including first-class gradients, just-in-time CUDA kernel compilation, automatic batching and support for new hardware such as TPUs.

This talk will demonstrate how Julia is increasingly becoming a natural language for machine learning, the kind of libraries and applications the Julia community is building, the contributions from India (there are many!), and our plans going forward.

There has been a lot of concerns about the black-box nature of AI. People have been asking for a sensible AI guideline with the weight of Law behind it. In April 2019, the EU released it's Ethics guidelines for trustworthy AI. Before that during the Obama administration, the National Science and Technology Council came up with its own set of broad guidelines called "Preparing for the Future of Artificial Intelligence."

Most of these cover an impressive amount of ground in several major categories:

• Transparency: Any time an AI system makes decisions on a user's behalf, that person should be aware of it. The reasoning behind decisions should be easily explainable.
• Safety: AI systems should be designed to withstand attempted hijacking and other attacks performed by hackers.
• Fairness: Decisions made by AI systems should not be influenced by gender, race or other personal identifiers. They should be as impartial as possible and not reflect human biases.
• Environmental stewardship: Not all the stakeholders in AI development are human. The development of these platforms and the implications of their decision-making and sustainability should take into account the needs of the larger environment and other forms of life.
• And so on...

At this conference, we would like to bring our experts together to hear their views/concerns on this topic.

This talk will help you understand the vision behind ODSC Conference and how it has grown over the years.

Today most Data Scientists focus on the art, science, and engineering of "Modelling" - how to build a model. But as AutoML is taking over, this skill is fast becoming obsolete.

In this talk, through a variety of examples, we will highlight an even more fundamental skill in Data Science: The Art of "Formulating" a specific Business problem, a Holistic Solution, or a Product feature as a Data Science problem.

In recent years, there has been a lot of research in the area of sequence to sequence learning with neural network models. These models are widely used for applications such as language modeling, translation, part of speech tagging, and automatic speech recognition. In this talk, we will give an overview of sequence to sequence learning, starting with a description of recurrent neural networks (RNNs) for language modeling. We will then explain some of the drawbacks of RNNs, such as their inability to handle input and output sequences of different lengths, and describe how encoder-decoder networks, and attention mechanisms solve these problems. We will close with some real-world examples, including how encoder-decoder networks are used at LinkedIn.

We will trace the journey of NLP over the past 50 odd years. We will cover chronologically Hidden Markov Models, Elman networks, Conditional Random Fields, LSTMs, Word2Vec, Encoder-Decoder models, Attention models, transfer learning in text and finally transformer architectures. Our emphasis is going to be on how the models became powerful and simple to implement simultaneously. To demonstrate this, we take a few case studies solved at INSOFE with a primary goal of retaining accuracy while simplifying engineering. Traditional methods will be compared and contrasted against modern models and show how the latest models actually are becoming easier to implement by the business. We also explain how this enhanced comfort with text data is paving way for state of the art inclusive architectures

When I was in grad school it was called statistics. A few years later I told people I did machine learning and after seeing the confused look on their face I changed that to data science which excited them. More years passed, and without changing anything I do, I now practice AI, which seems scary to some people and somehow involves ML. During this talk we will demystify buzzwords, technical terms and overarching ideas. We'll touch upon key concepts and see a little bit of code in action to get a sense of what is happening in ML, AI or whatever else we want to call the field.

Within Nokia Software Digital Experience, we build products that increase customer satisfaction and reduce churn through proactive identification of the user problems and that enable service providers to resolve problems faster. To achieve such tasks, ML and DL techniques are now contributing a lot to these successes. However, there is usually a long journey between building a first model up-to delivering a field-proven product. Besides providing highlights on how machine and deep learning are used today to boost the broadband connection, this talk will reveal some challenges encountered and best-practices involved to reach the expected quality level.

In the last few years, when the cybercrooks have speeded their execution plan on making quick money by ransomware attacks. All enterprises, including banks, government offices, police stations, big and small businesses, have witnessed WannaCry, Petya, NotPetya ransomware attacks. The question for us is what we can do to defend from cyber threats? The cybersecurity industry is pitching heavily to leverage AI to combat cyber threats. Almost every cybersecurity vendor is claiming to have AI in its product. This makes it difficult for end-user enterprises to choose the product, and they need to evaluate the AI capabilities of multiple vendors. In this talk, I will cut the hype and discuss the reality of what AI can do for cybersecurity? I will share use cases, data pipeline, architecture, algorithms that are proven for information security along with the challenges in deploying them in the wild. The audience will be able to learn how to combine AI with domain knowledge to make an enterprise AI solution.

In this workshop, we will understand the algorithms behind recommender systems in different domains and gain an appreciation for how the domain impacts the approach used. Attendees will be creating recommenders using user item matrices, news and graphs gaining an understanding of collaborative and content-based filtering, text representation, matrix factorization, and random walks.

Modern machine learning (ML) offers a new way of creating software to solve problems, focused on learning structures, learning algorithms, and data. In all steps of this process, from the specification of the problem, to the datasets chosen as relevant to the solution, to the choice of learning structures and algorithms, a variety of biases can creep in and compound each other. In this talk, we present a systems view of detecting Bias in AI/ML systems as analogous to the software testing problem. To start, a variety of expectations from an AI/ML system can be specified given its intended goals and deployment. Different kinds of bias can then be mapped to different failure modes, which can then be tested for during a variety of techniques. We will also describe a new technique based on Topological Data Analysis to detect bias in source datasets. This technique utilizes a persistence homology based visualization and is lightweight: the human-in-the-loop does not need to select metrics or tune parameters, and carry out this step before choosing a model. We’ll describe experiments on the German credit dataset using this technique to demonstrate its effectiveness.

Dr.Vikram Vij, Senior Vice President, Head of Voice Intelligence Team, Samsung Research India – Bangalore (SRIB) will share the journey that Samsung has undertaken in developing its Voice Assistant Bixby and particularly Automatic Speech Recognition(ASR) system that powers it. ASR is one of the complex engines that power modern virtual Assistants. Several independent components such as pre-processors (Acoustic Echo Cancellation, Noise Suppression, Neural Beam forming and so on), Wake word detectors, End-point detectors, Hybrid Decoders, Inverse Text Normalizers work together to make a complete ASR system. We are in an exciting period with tremendous advancements made in recent times. The development of End-to-End(E2E) ASR systems is one such advancement that has boosted recognition accuracy significantly and it has the potential to make speech recognition ubiquitous by fitting completely On-Device thereby bringing down the latency and cost and addressing the privacy concerns of the users. Samsung, the largest device maker on the planet, envisions a huge value in bringing Bixby to a variety of existing devices and new devices such as Social Robots, which throws many technical challenges particularly in making the ASR very robust. In this talk, Dr.Vikram is excited to present the cutting-edge technologies that his team is developing - Far-Field Speech Recognition, E2E ASR, Whisper Detection, Contextual End-Point Detection (EPD), On-device ASR and so on. He would also elaborate on the research activities his team is relentlessly pursuing.

Modern statistics has become almost synonymous with machine learning, a collection of techniques that utilize today's incredible computing power. This two-part course focuses on the available methods for implementing machine learning algorithms in R, and will examine some of the underlying theory behind the curtain. We start with the foundation of it all, the linear model. We look how to assess model quality with traditional measures and cross-validation and visualize models with coefficient plots. Next we turn to penalized regression with the Elastic Net. After that we turn to Boosted Decision Trees utilizing xgboost. Along the way we learn modern techniques for preprocessing data.

“Is this the only data you have?” An expression of surprise not uncommonly encountered when evaluating a new opportunity to apply data science. Suitability of available data is a key factor in the abandonment of many otherwise well considered data science initiatives.

"Could the folks who were responsible for the design of the business process and the supporting IT applications not been more forward thinking and captured the more of the relevant data? To make it even worse, for the data that is being captured, the manual entries are not even consistent between the operators."

Well, don't throw up you hands just yet. If you are a relatively newly minted data scientist, you are probably used to data being served to you on a platter! (Kaggle, UCI, Imagenet ..add your favourite platter to the list)

Generally 3 types of challenges are present..

• At one extreme.. They are building a new app. They want to incorporate a recommendation engine. The app is not released ! There is no data, zero, nada, zilch..
• At the other extreme.. I want us to build a up-sell engine. They have a massive database with a huge number of tables. If I just look for revenue related fields, I see 10 different customer revenue fields! Which is the right one to use!!
• The client wants me to build a promotion targeting engine. But they keep changing their offers every month! By the time I have enough data for a promotion, they are ready to kill that promotion move on to some other promotion..
• They want to build a decision support engine. But the available attributes capture only 20-30% of what goes into making the decision. How it this going to be of any help?

Sounds familiar? You are not alone. The speaker using case studies from his own experience will guide the audience on how they can make the best of the situation and deliver a value adding data science solution, or how to decide whether it is more prudent to not pursue it after all.

We will train the audience how to develop advanced image segmentation with FCN/DeepLab algorithms which can help visualize the driving scenarios accurately, so as to allow the autonomous driving system to take appropriate action considering the obstacle views.

One of the principal problems in the developing world is the poor localization of its addresses. This inhibits discoverability of local trade, reduces availability of amenities such as creation of bank accounts and delivery of goods and services (e.g., e-commerce) and delays emergency services such as fire brigades and ambulances. In general, people in the developing World identify an address based on neighbourhood/locality names and points of interest (POIs), which are neither standardized nor any official records exist that can help in locating them systematically. In this paper, we describe an approach to build accurate geographical boundaries (polygons) for such localities.

As training data, we are provided with two pieces of information for millions of address records: (i) a geocode, which is captured by a human for the given address, (ii) set of localities present in that address. The latter is determined either by manual tagging or by using an algorithm which is able to take a raw address string as input and output meaningful locality information present in that address. For example, for the address, “A-161 Raheja Atlantis Sector 31 Gurgaon 122002”, its geocode is given as (28.452800, 77.045903), and the set of localities present in that address is given as (Raheja Atlantis, Sector 31, Gurgaon, Pin-code 122002). Development of this algorithm are part of any other project we are working on; details about the same can be found here.

Many industries, such as the food-delivery industry, courier-delivery industry, KYC (know-your-customer) data-collection industry, are likely to have huge amounts of such data. Such crowdsourced data usually contain large a amount of noise, acquired either due to machine/human error in capturing the geocode, or due to error in identifying the correct set of localities from a poorly written address. For example, for the address, “Plot 1000, Sector 31 opposite Sector 40 road, Gurgaon 122002”, a machine may output the set of localities present in this address as (Sector 31, Sector 40, Gurgaon, Pin-code 122002), even though it is clear that the address does not lie in Sector 40.

The solution described in this paper is expected to consume the provided data and output polygons for each of the localities identified in the address data. We assume that the localities for which we must build polygons are non-overlapping, e.g., this assumption is true for pin-codes. The problem is solved in two phases.

In the first phase, we separate the noisy points from the points that lie within a locality. This is done by formulating the problem as a non-linear multi-classification problem. The latitudes and longitudes of all non-overlapping localities act as features, and their corresponding locality name acts as a label, in the training data. The classifier is expected to partition the 2D space containing the latitudes and longitudes of the union of all non-overlapping localities into disjoint regions corresponding to each locality. These partitions are defined as non-linear boundaries, which are obtained by optimizing for two objectives: (i) the area enclosed by the boundaries should maximize the number of points of the corresponding locality and minimize the number of points of other localities, (ii) the separation boundary should be smooth. We compare two algorithms, decision trees and neural networks for creating such partitions.

In the second phase, we extract all the points that satisfy the partition constraints, i.e., lie within the boundary of a locality L, as candidate points, for generating the polygon for locality L. The resulting polygon must contain all candidate points and should have the minimum possible area while maintaining the smoothness of the polygon boundary. This objective can be achieved by algorithms such as concave hull. However, since localities are always bounded by roads, we have further enhanced our locality polygons by leveraging open source data of road networks. To achieve this, we solve a non-linear optimisation problem which decides the set of roads to be selected, so that the enclosed area is minimized, while ensuring that all the candidate points lie within the enclosed area. The output of this optimisation problem is a set of roads, which represents the boundary of a locality L.

The field of Artificial Intelligence powered by Machine Learning and Deep Learning has gone through some phenomenal changes over the last decade. Starting off as just a pure academic and research-oriented domain, we have seen widespread industry adoption across diverse domains including retail, technology, healthcare, science and many more. More than often, the standard toolbox of machine learning, statistical or deep learning models remain the same. New models do come into existence like Capsule Networks, but industry adoption of the same usually takes several years. Hence, in the industry, the main focus of data science or machine learning is more ‘applied’ rather than theoretical and effective application of these models on the right data to solve complex real-world problems is of paramount importance.

A machine learning or deep learning model by itself consists of an algorithm which tries to learn latent patterns and relationships from data without hard-coding fixed rules. Hence, explaining how a model works to the business always poses its own set of challenges. There are some domains in the industry especially in the world of finance like insurance or banking where data scientists often end up having to use more traditional machine learning models (linear or tree-based). The reason being that model interpretability is very important for the business to explain each and every decision being taken by the model.However, this often leads to a sacrifice in performance. This is where complex models like ensembles and neural networks typically give us better and more accurate performance (since true relationships are rarely linear in nature).We, however, end up being unable to have proper interpretations for model decisions.

To address and talk about these gaps, I will take a conceptual yet hands-on approach where we will explore some of these challenges in-depth about explainable artificial intelligence (XAI) and human interpretable machine learning and even showcase with some examples using state-of-the-art model interpretation frameworks in Python!

It is too tedious to keep on asking questions, seek explanations or set thresholds for trends or anomalies. Why not find problems before they happen, find explanations for the glitches and suggest shortest paths to fixing them? Businesses are always changing along with their competitive environment and processes. No static model can handle that. Using dynamic models that find time-delayed interactions between multiple time series, we need to make proactive forecasts of anomalous trends of risks and opportunities in operations, sales, revenue and personnel, based on multiple factors influencing each other over time. We need to know how to set what is “normal” and determine when the business processes from six months ago do not apply any more, or only applies to 35% of the cases today, while explaining the causes of risk and sources of opportunity, their relative directions and magnitude, in the context of the decision-making and transactional applications, using state-of-the-art techniques.

Real world processes and businesses keeps changing, with one moving part changing another over time. Can we capture these changing relationships? Can we use multiple variables to find risks on key interesting ones? We will take a fun journey culminating in the most recent developments in the field. What methods work well and which break? What can we use in practice?

For instance, we can show a CEO that they would miss their revenue target by over 6% for the quarter, and tell us why i.e. in what ways has their business changed over the last year. Then we provide the prioritized ordered lists of quickest, cheapest and least risky paths to help turn them over the tide, with estimates of relative costs and expected probability of success.

At idealo.de we store and display millions of images. Our gallery contains pictures of all sorts. You’ll find there vacuum cleaners, bike helmets as well as hotel rooms. Working with huge volume of images brings some challenges: How to organize the galleries? What exactly is in there? Do we actually need all of it?

To tackle these problems you first need to label all the pictures. In 2018 our Data Science team completed four projects in the area of image classification. In 2019 there were many more to come. Therefore, we decided to automate this process by creating a software we called Image ATM (Automated Tagging Machine). With the help of transfer learning, Image ATM enables the user to train a Deep Learning model without knowledge or experience in the area of Machine Learning. All you need is data and spare couple of minutes!

In this talk we will discuss the state-of-art technologies available for image classification and present Image ATM in the context of these technologies. We will then give a crash course of our product where we will guide you through different ways of using it - in shell, on Jupyter Notebook and on the Cloud. We will also talk about our roadmap for Image ATM.