Empowering Diversity in Technology: Women in AI Dinner, Houston

Original

More and more, companies are recognising the potential of women to be equal to their male counterparts in all areas of technology, including AI. This is partly thanks to the obstacles that women in top roles such as Daphne Koller, Fei Fei Li and Joelle Pineau have overcome to achieve these positions and destigmatise the roles, but more can be done to improve the gender disparity in management and leadership positions.

As we bring our global summits to new locations, so do we expand the RE•WORK Women in AI Dinners to showcase top women working in the industry as well as bringing together some of the most promising talent from the next generation keen to make their mark in the rapidly advancing space. Tonight in Houston, before sitting down to a three-course dinner and keynote presentations from AI experts, attendees mingled over a champagne reception and got to know each other. At this evening’s event, kindly sponsored by Capital One, we saw first hand how AI is impacting every industry and this was demonstrated by the diverse range of attendees representing fields such as Healthcare, Media and Technology, Financial Services, Natural Resources, Fashion & Retail, Education, Government and more.

"Homogeneity is the killer of innovation. AI can change the world for the better, but unless we are all integrating diverse perspectives, backgrounds and experiences at every level of our technology development, machine learning advancements are only going benefit a small segment of the world's population -- likely the privileged few. However, it's not just a social imperative. Diversity on the development side opens doors for new markets, and, ultimately, that means higher sales and revenue. It's just good business sense." - Courtenay Siegfried, Founding Partner at Alice (Dinner Compere)


Courtenay kicked off the evening by reminding the attendees that ‘diversity in AI is a two-way street.’ It's about attracting and training diverse talent, but it's also about ensuring the product works for diverse end users. Alice is a free website that helps business owners find the right path to start and grow their companies and supports equal access and opportunity for all business owners, no matter where they come from or who they know. Founders want connections, mentorship, and to know how to get funding. Alice wanted to change the world and realised they had to turn to technology and machine learning to facilitate this. ‘At Alice, we know that our technology is only as good as the data we put into it, so it's critically important to make sure the data truly represents the diversity of our users. We also make sure to test against diverse audiences, so we know what works and doesn't work for business owners from all backgrounds and experiences. It's incredibly easy to inadvertently introduce unintentional biases into machine learning algorithms, so all of us need to be active in checking our assumptions.’

First up to share her most recent work was Margaret Mayer, Vice President of Software Engineering at Capital One. Margaert discussed how convolutional neural networks and long short-term memory networks are enabling more sophisticated NLP systems and building the necessary components to lay the foundation for advanced dialogue. She began by explaining how her Mom dropped out of an engineering course at College and went on to get an English degree. When it came to Margaret going to college, her Mom could have easily discouraged a technical path because of her experience, but she said ‘to go for it without hesitation’. Now working on Eno at Capital One, she explained that ‘the future is conversation and the future is voice. There’s something deeper in conversation than a distinct question and answer which is what we’re trying to figure out with our bot - we want a connection.’ Capital One want their bot to be seen as intelligent, ‘so we need to work out when we want to ask to clarify and also inject personality to try and create a conversation.’ Margaret used an example of the two-way fraud alerts that have been in place for a long time and explained how they’re working to improve these: ‘For example, if I’m in Houston and someone’s using my credit card in Florida - maybe it’s me, maybe it’s not - our algorithm decides if it’s fraudulent or not. It sends you an SMS saying ‘is it you, confirm or deny’. This might seem like a straightforward question, but about a third of people reply with ‘maybe, don’t know, sort of’ but this response isn’t part of the algorithm. With our bot, Eno, we used natural language processing to classify and identify what the customer is saying so we can help them further. For us, this is a way that Eno is providing tangible value to our users.’

With such diverse guests at the RE•WORK dinners, we understand the importance of attendees having the opportunity to chat with as many experts as possible. To help facilitate these connections, between each course we encouraged everyone to get up and move seats, introduce themselves to their new neighbour and start new conversations. Once everyone had moved around and met their new table, Giewee Hammond, Lead Data Scientist, Aramco Services shared her work on The Sparse Data Problem in Machine Learning. Giweee opened up the conversation to the audience and asked everyone to ‘help us understand bad data’, and we received several responses, including ‘bad data is any obsolete old data that’s inconsistent, not timely, irrelevant, one of the major issues I’ve seen is inconsistent data leading to inaccuracies’. Giewee went on to explain that the type of bad data she is referring to is missing or sparse data. For example, both types of data have the same effect on data that a concave mirror might have on the human figure. ‘Sparse and missing data can have effects on data to make it appear like it’s fulfilling a certain function when in fact it is not. Information bias is a common problem, and there’s a way to solve this. Information bias is collecting the wrong data and interpreting it incorrectly.’ Aramco Services are looking at ways to increase model accuracy with sparse data through multiple amputation through deep learning, where it compensates for both sparse an missing data, allowing them to come up with really effective results. When applying deep learning to data, if the data is over 65% sparse data, deep learning improves it massively. You can get away with using deep learning tools on sparse data where you might not get away with traditional machine learning models. 


To round off the evening we heard from Rupa Kanchi, Postdoctoral Fellow, MD Anderson Cancer Center who spoke about Machine Learning in Medical Imaging. Of course we’ve heard many cases for the application of AI for Social Good this evening, and healthcare is undoubtedly one of the industries that has the opportunity to be positively transformed the most. Rupa’s research focuses on the use of bioinformatics and machine learning methods in order to understand the diverse data types in cancer genetics. ‘The different imaging modalities (ultrasound, CT, MRI, etc.) are playing an increasingly important role in disease detection, diagnosis and treatment, and substantial developments have happened in data acquisition, management, sharing and ownership, and adoption of a standardized language in writing imaging reports.’ Rupa explained how developments in image analysis are still in their early stages, and machine learning can help here as an applied branch of AI. ’ML presents opportunities for quality improvement in image data analysis in all stages of radiotherapy. Some of the reasons for the recent popularity of machine learning are the increasing availability of large image data sets and GPU-based computing capabilities, as well as a snowballing community of deep learning algorithm creators. ‘Our lab is directed towards research in understanding and implementing deep learning methods in the area of image segmentation and classification’. Rupa closed by sharing how researchers and radiologists often work with series’ of images and deep learning can speed up the process of identifying the targets in these images. ‘Automating this process can enormously increase productivity of radiologists’.’

As the night came to a close, Courtenay followed on the topic of using AI for good by reminding us that 'the potential for AI to create and facilitate positive social good and solve global problems is nearly endless (when it's done consciously and with forethought). Machine learning can and is being used to identify areas being impacted by poverty, more efficient agriculture practices, improve healthcare, and break down knowledge barriers. It's a really exciting time.'

The RE•WORK team is ending 2018 in Houston with the Applied AI Summit and Machine Learning for DevOps Summit taking place this Thursday and Friday. Can’t make it to Houston but keen to get involved? We’re all excited to return to San Francisco next January 2019 for the world’s biggest Deep Learning Summit, as well as the Women in AI Dinner. We’re currently offering 25% off all Summit passes when you register before Friday 30th with the code CYBER25.

Original

Machine Learning AI Deep Learning Summit International Women's Day Women in Tech


0 Comments

    As Featured In

    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original
    Original

    Partners & Attendees

    Intel.001
    Nvidia.001
    Acc1.001
    Ibm watson health 3.001
    Facebook.001
    Rbc research.001
    Mit tech review.001
    Graphcoreai.001
    Maluuba 2017.001
    Twentybn.001
    Forbes.001
    Kd nuggets.001
    This website uses cookies to ensure you get the best experience. Learn more