Обзор конференций по анализу изображений и распознаванию образов

19 окт 2011 11:25 19 окт 2011 11:39 от Lex.
Lex
Модератор
Модератор
Сообщений: 4048
Больше
Обзор конференций по анализу изображений и распознаванию образов #14097

aleksandr пишет: Не прибедняйтесь: сами-то себе можете объяснить...
Что хаос в картинке - конечной сложности...

Вы пишете: "...более чем достаточно..." - а ДЛЯ ЧЕГО достаточно?

Для картинки 1000x1000 это 2^ 24000000. Неопределенности такой размерности достаточно для любой задачи к которой мы вообще способны прикоснуться. Последовательности в пространстве такой размерности вполне можно считать хаотическими.

Для "распознавания образа" достаточно?

Для того чтобы наша система считалась отрытой. Для этого хаоса в картинках вполне достаточно.

Книжки почитать советуете?
Не написана ещё эта книжка, пишу ещё...

Чукча не читатель? ))))))

И у нас совсем совсем нет книжек о том как отделить структуру сообщения от неопределенности ключа в дампе? Ха ха! ))

Хмур пишет: Вы не понимаете самой сути. Нет универсального метода. Чем универсальней тем менее
эффективней. Универсальный метод — это именно поиск невозможного, и именно для такого
поиска нужен ваш лжекреатив с косами и навями.. :)

Как гипотеза о том что слова есть случные наборы звуков которые формировались отдельно от культурного пространства доказывает отсутствие универсального алгоритма для распознавания образа? ))))))

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

19 окт 2011 11:32 19 окт 2011 11:34 от Lex.
Lex
Модератор
Модератор
Сообщений: 4048
Больше
Обзор конференций по анализу изображений и распознаванию образов #14098
Вот настоящий вопрос. Как формально доказать что обе картинки реализуют один и тот же образ? Что в этом случае такое образ? Не думаю что эвристики и "уровни" нам тут помощники.







Вложения:

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

19 окт 2011 13:18
NO.
Академик
Академик
Сообщений: 2213
Больше
Обзор конференций по анализу изображений и распознаванию образов #14099
По двум картинкам никак, нужны ещё.
Например картина "Боярыня Морозова", по сравнению с ней эти две похожи.

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

19 окт 2011 14:43
Lex
Модератор
Модератор
Сообщений: 4048
Больше
Обзор конференций по анализу изображений и распознаванию образов #14101
Что две что сто. Все равно нужно сформовать вектор признаков.

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

19 окт 2011 15:02 19 окт 2011 15:06 от NO..
NO.
Академик
Академик
Сообщений: 2213
Больше
Обзор конференций по анализу изображений и распознаванию образов #14103
В каком пространстве вектор? В пространстве сюжетов по килобайту текста признаков будет 2^8000 штук, такой вектор ни в какой винчестер не влезет, терабайт это всего вектор 2^43 бинарных признаков.

Долой статистику! Смерть старухе-процентщице!

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

19 окт 2011 16:32
Lex
Модератор
Модератор
Сообщений: 4048
Больше
Обзор конференций по анализу изображений и распознаванию образов #14104
а для картинки 1000x1000 2^24000000 ))))))) Короче нужны алгоритмы отсева ботвы и тут все равно сколько у вас картинок, в любом случае при делении их количества на 2^24000000 будет ноль )))

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

19 окт 2011 17:21 19 окт 2011 17:30 от NO..
NO.
Академик
Академик
Сообщений: 2213
Больше
Обзор конференций по анализу изображений и распознаванию образов #14106
Я думаю ценными будут технологии, которые делают то, что не могут люди.
1. С высокой скоростью выполнять простейшее распознавание вроде индексации видео или поиска лиц террористов в реальном времени.
2. Медленный анализ сложных изображений. Вроде рентгеновских снимков или по следам на земле разобрать какие там ездили машины и что делали.

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

19 окт 2011 17:31
NO.
Академик
Академик
Сообщений: 2213
Больше
Обзор конференций по анализу изображений и распознаванию образов #14107
Вложения:

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

20 окт 2011 04:31
aleksandr
Академик
Академик
Сообщений: 2802
Больше
Обзор конференций по анализу изображений и распознаванию образов #14115
2Хмур:
Всё, что Вы мне написали, мне понятно (как я считаю).
На моей новой модели это просто и понятно.
Так просто и ясно, что мои квадратики и стрелочки будут во всех учебниках, и довольно скоро.

Но...
Я не просто так написал про оксюморон хаотического сигнала.
Сигнал однороден хаосу (динамическому) - но это не сам динамический хаос.
Да - я согласен, что хаос - носитель информации, и я согласен с Лексом, что это основано на родовом свойстве всех хаосов - сложности.
Сигнал - одно из лиц информации, которая ну никак не хаос - но без хаоса никак!

"Наблюдатель" - отдельная песня, уже в принципе спетая квантовыми представлениями.
Но у меня это выглядит значительно проще, чем у Шрёдингера, и без налёта мистики, как у копенгагенских интертрепаторов.

2Лекс:
Я на вашей стороне - универсальный метод нужен!
Но - это не универсальный алгоритм!
Алгоритм создаётся под каждое распознавание - специфический.
Вы же отлично начали - "теорией о восприятии", синтез "приёмника" - это первый этап синтеза алгоритма!
Продолжайте!

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

20 окт 2011 04:37
aleksandr
Академик
Академик
Сообщений: 2802
Больше
Обзор конференций по анализу изображений и распознаванию образов #14116

Хмур пишет: Александро>

Я понимаю, когда из хорошо кодированного сигнала как хаоса мы извлекаем информацион-
ную структуру, то мы ее интерпретируем путем соотнесения с др. структурой (отображаем
одно в другое), а это другое в своей системе кодирования тоже может быть представлено как
хаотический сигнал. У вас в голове замкнуло эту интуицию, и вы стали не так давно бегать
с идеей извлекать информацию 'хаосом из хаоса'. Может быть сейчас вы поймете, что такое
понимание — это фикция, это внешняя ассоциация, это потеря конструктивных, главных,
промежуточных этапов. Хаос — это только носитель информации.. (и носитель, и способ
модуляции и способ кодирования)....


Нет, я так не думал, Вы меня неправильно поняли!
Ваше "..а это другое в своей системе кодирования..." нисколько не хаотическое, а детерминированное до усирачки!
Тут бы надо и Лексу репку поднапрячь: он считает, что хаоса в картинке ему достаточно (т.е. сигнала).
А я говорю - нет, недостаточно!
Нужен другой хаос - бесконечной сложности, и вот тогда, "хаосом по хаосу", ...

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

20 окт 2011 09:50
Lex
Модератор
Модератор
Сообщений: 4048
Больше
Обзор конференций по анализу изображений и распознаванию образов #14119

aleksandr пишет: Я на вашей стороне - универсальный метод нужен!
Но - это не универсальный алгоритм!
Алгоритм создаётся под каждое распознавание - специфический.

Будет универсальный метод, будет и универсальный алгоритм.

NO.
Вы себе представляете на каком пространстве решений должен стоять алгоритм, что бы на рентгеновском снимке найти Гомэра Симпсона? ))) Весь интернет проиндексируем? )))

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

20 окт 2011 12:20
NO.
Академик
Академик
Сообщений: 2213
Больше
Обзор конференций по анализу изображений и распознаванию образов #14123
Пространство решений вроде "Гомер Симпсон позирует гламурному фотографу с красным яблоком в левой руке" ещё больше.

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

21 окт 2011 04:15
aleksandr
Академик
Академик
Сообщений: 2802
Больше
Обзор конференций по анализу изображений и распознаванию образов #14125

Хмур пишет: Александро>

> оксюморон "...хаотический сигнал...

Это не жаренный лед. Это строгое понятие. Динамичеаский хаос — это сложная структура.
Это не белый шум, не цветные производные белого шума — это носитель информации (раз
это сложнострутурированный объект). Cтороннему наблюдателю хаотический сигнал
может казаться статистическим хаосом, шумом, случайным процессом Винера и т.п. -
но адекватному приемнику (я ж рассказывал вам про наблюдателей, про синхронизацию
хаосов
) ХС выдаст информацию, закодированную в его структуре....


Приёмник - это ещё не наблюдатель, хаосы не синхронизированы.
А наблюдатель сиречь синхронизированные хаосы - уже не приёмник, он, как Вам прекрасно известно, самим наблюдением меняет информацию...
Такая вот копень-гагень тирпертация....

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

21 окт 2011 04:19
aleksandr
Академик
Академик
Сообщений: 2802
Больше
Обзор конференций по анализу изображений и распознаванию образов #14126

Lex пишет:

aleksandr пишет: .............

Книжки почитать советуете?
Не написана ещё эта книжка, пишу ещё...

Чукча не читатель? ))))))

И у нас совсем совсем нет книжек о том как отделить структуру сообщения от неопределенности ключа в дампе? Ха ха! )) .....


А там написано, чем становится, во что превращается образ распознанный?
Вот когда напишете, тогда смейтесь, это будет хороший смех...

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

24 сен 2012 05:45
NO.
Академик
Академик
Сообщений: 2213
Больше
Обзор конференций по анализу изображений и распознаванию образов #22195
Собирать грибы поможет смартфон
www.vesti.ru/doc.html?id=914459

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

24 сен 2012 19:17
slava
Старожил
Старожил
Сообщений: 1205
Больше
Обзор конференций по анализу изображений и распознаванию образов #22223
See Who's Attending PAW Boston
Gregory Piatetsky, KDnuggets Editor Этот адрес электронной почты защищён от спам-ботов. У вас должен быть включен JavaScript для просмотра. via in00.m1e.net

Predictive Analytics World 2012 Boston

Less Than 2 Weeks to Register for
Predictive Analytics World Boston, September 30 - October 4, 2012

Join us and network with predictive analytics experts and thought leaders who come to PAW to share their latest techniques and insight. PAW's comprehensive agenda provides case studies from fortune 1000 companies as well as in-depth workshops from the some of the most respected and sought-after experts in the industry.

Keynotes and plenary session speakers include:

Scott Nicholson, Chief Data Scientist, Accretive Health (formerly of LinkedIn)
Beyond Big Data

Anne Robinson, Dir. of Supply Chain Strategy & Analytics, Verizon Wireless
Influencers, Skeptics, and Data Geeks

Robert Jewell, Dir. Global Bus Dev & Partnerships, IBM Watson Solutions
Putting IBM Watson to Work

Marc Smith, Chief Social Scientist, Connected Action Consulting Group
Charting Collections of Connections in Social Media

John Elder, CEO & Founder, Elder Research, Inc.
Becoming an Ace with a Robot as your Wingman!


www.predictiveanalyticsworld.com/boston/...nt=09-20-12#day2-130

=========================================

Full Agenda – Boston 2012
Agenda Overview | Full Agenda | Speakers | Register me! | Day 2
All level tracks Track 1 sessions are for all levels.
Track 2 sessions are expert/practitioner level.
Conference Day 1: Monday, October 1, 2012

7:30-9:00am

Registration & Breakfast

[ Top of this page ] [ Agenda overview ]

9:00-9:45am

Keynote
Case Study: LinkedIn and Accretive Health
Beyond Big Data: Better Living Through Data Science

The 'big data' theme is overrated and can be misleading. Ultimately data and data analytics cannot get you completely over the finish line; you also need a combination of asking the right questions, context/product intuition, and in some cases an understanding of the psychology behind decision-making. This end-to-end ownership and expertise are the role of the data scientist, and help your big/huge/fat data achieve the inflection point that leads to big insights. Using lessons from consumer internet (LinkedIn and online advertising), health care analytics (Accretive Health) and behavioral economics, we will discuss examples of how the combination of data science and different representations of 'big data' generate insights that help people make better decisions about their lives.

Speaker: Scott Nicholson, Chief Data Scientist, Accretive Health (formerly of LinkedIn)

[ Top of this page ] [ Agenda overview ]

IBM 9:45-10:05am

Gold Sponsor Presentation
Raising the Bar for Predictive Analytics Deployment: The Newest Techniques

Although the use of predictive analytics has come a long way in recent years, it is clear that there are now much higher expectations for wider and more accurate deployment methods. So while more organizations see the value of analytics, few are comfortable with their current tools and abilities to create and deploy useful solutions. In this session we'll explore the very newest techniques and capabilities that have emerged to help you ingrain predictive analytics into the DNA of your organization, and deploy solutions that empower your team to make the right decisions and consistently deliver the best results.
Speakers: Jason Verlen, Director, SPSS Product Strategy & Management, IBM Software Group

[ Top of this page ] [ Agenda overview ]
JMP

10:05-10:15am

Gold Sponsor Presentation
Modeling Renewal Probability In Order to Boost It at Tripadvisor

For any subscription–based business, there is the danger that the subscription will not be renewed at the end of its term. Modeling the probability of renewal is a necessary, but not sufficient first step towards improving the likelihood of renewal. At Tripadvisor for Business, we use the rate of change in renewal probability as model parameters are varied to identify subscriptions that will benefit most from a marketing intervention.

Speaker: Michael Berry, Analytics Director, TripAdvisor

[ Top of this page ] [ Agenda overview ]

10:15-10:40am

Breaks / Exhibits

[ Top of this page ] [ Agenda overview ]

10:40-11:25am

All level tracks Track 1: Thought Leadership
The Practical Data Scientist

The term Data Science has been derided as hype wrapped around the traditional role of the analyst. But if we look at the type of data we have today, the software we use to manipulate it, and the wide range of statistics, machine learning, and analytics tools at our disposal, it is fair to say that we are in a new world. Based on more than 20 interviews with top data scientists and a variety of other research, Dan Woods, CTO and Editor of CITOResearch.com and a contributor to Forbes.com, defines a practical approach to data science, one focused on experimentation, gradual development of skills, and achieving business value. Woods defines a reference model for data science, explores various types of maturity models, suggests organizational structures, and reviews technology that allows for a quick start without a large budget.
Speaker: Dan Woods, CTO, Chief Editor/Analyst and Founder, Evolved Media, and a Forbes Contributor
10:40-11:25am

Track 2: HR Analytics
Case Study: U.S. Special Forces
Hiring and Selecting Key Personnel Using Predictive Analytics

Hiring and selection of personnel in specialized work environments incurs huge direct and opportunity costs for organizations. One of the largest challenges is that the selection process is often left in the hands of those with either high experience in the domain area but little experience in selection or vice versa.

Predictive Analytics and statistics can play a critical role in formalizing and automating much of the selection process. This session provides an overview of the selection processes using both measures of skills and psychological measures to quantify IQ, domain knowledge, grit, and determination. Examples will be drawn from hiring practices for Special Forces (such as Army Rangers and Navy SEALs) and predictive analytics teams.

Speaker: Dean Abbott, President, Abbott Analytics, Inc.

[ Top of this page ] [ Agenda overview ]
11:30am-12:15pm

Keynote
Charting Collections of Connections in Social Media: Creating Maps and Measures with NodeXL

Networks are a data structure common found across all social media services that allow populations to author collections of connections. The Social Media Research Foundation's NodeXL project makes analysis of social media networks accessible to most users of the Excel spreadsheet application. With NodeXL, Networks become as easy to create as pie charts. Applying the tool to a range of social media networks has already revealed the variations present in online social spaces. A review of the tool and images of Twitter, flickr, YouTube, and email networks will be presented.

We now live in a sea of tweets, posts, blogs, and updates coming from a significant fraction of the people in the connected world. Our personal and professional relationships are now made up as much of texts, emails, phone calls, photos, videos, documents, slides, and game play as by face-to-face interactions. Social media can be a bewildering stream of comments, a daunting fire hose of content. With better tools and a few key concepts from the social sciences, the social media swarm of favorites, comments, tags, likes, ratings, and links can be brought into clearer focus to reveal key people, topics and sub-communities. As more social interactions move through machine-readable data sets new insights and illustrations of human relationships and organizations become possible. But new forms of data require new tools to collect, analyze, and communicate insights.

Speaker: Marc Smith, Chief Social Scientist, Connected Action Consulting Group

[ Top of this page ] [ Agenda overview ]

Forsee

12:15-12:30pm

Platinum Sponsor Presentation
Managing Forward: Analytics For Today's Multi-Channel, Multi-Device Consumer

When done right, customer satisfaction measurement can yield more than just insights into how well your company, brand, or channel (e.g., web, mobile, store) is performing today. It can also predict the likelihood of customers to engage in critical future behaviors. However, not all methodologies are created equal. They must answer three essential questions of management while demonstrating success not only in theory but in the marketplace.

Speaker: Eric Feinberg, Senior Director of Mobile, Media and Entertainment, ForeSee

[ Top of this page ] [ Agenda overview ]

12:30-12:45pm

Lightning Round of 2-Minute Sponsor Presentations

FICO Statsoft Salford Systems UCI Pervasive Decision Systems Mu Sigma

[ Top of this page ] [ Agenda overview ]

12:45-1:40pm

Lunch / Exhibits

[ Top of this page ] [ Agenda overview ]

1:40-2:25pm

Keynote
Case Study: Verizon
Influencers, Skeptics, and Data Geeks: Using Analytics to Drive Organizational Change

"What gets measured gets done" often is true when it comes to tactical execution. When applied to large-scale strategy, however, the implications of this adage are even more significant.

Advanced analytics can help reveal the true performance drivers in an organization. By leveraging the power of analytics in combination with the principles of change management, learn how to effectively lead your organization into a new era of operational success.

Speaker: Anne G. Robinson, Dir. of Supply Chain Strategy & Analytics, Verizon Wireless

[ Top of this page ] [ Agenda overview ]

Lattice Logo
2:25-2:55pm

Platinum Sponsor Presentation
How Big Data Delivers a Competitive Advantage

There's a lot of discussion as to whether Big Data can live up to its hype. Apply it to a businesses' front line – sales - and the benefits become tangible.

Personalized and predictive selling is one of Big Data's more interesting applications. From Amazon's Recommendations to Google's targeted ads, we are all being touched by Big Data whether or not we know or like it. While the power of Big Data is recognizable on the retail side, what is less visible is its application to B2B sales. From a sales rep's perspective, Big Data answers a major question: how do I find the customers who are most receptive to my product or service at a given time? The best reps already have a talent for this. Big Data democratizes this excellence by automating the skills of excellent reps by delivering insight out of massive amounts of internal, external and social data.

This session will discuss the five steps to generate insight from Big Data, using customer case studies as an example of how to apply predictive analytics to selling complex product and service offerings to gain competitive advantage.

Speaker: Kent McCormick, Ph.D., President & CTO, Lattice Engines

[ Top of this page ] [ Agenda overview ]

3:00-3:45pm
All level tracks Track 1: Big Data
Big Data and Big Analytics Trends: The Promise and the Hype

We will look at the current trends and buzzwords in Big Data, Data Mining and Predictive analytics field and examine how much hype and reality is in the promise of Big Data. We also analyze the growing and changing demand for data scientist skills and see which skills are the hottest.

Speaker: Gregory Piatetsky-Shapiro, Editor, KDNuggets
3:00-3:45pm

Track 2: Market Mix Optimization
Case Study: Penske
Marketing Mix Optimization: Forecasting and decision making under uncertainty

Most marketing mix solutions either fall short either due to reliance on imperfect information or failing to take into account management assessments and business uncertainties. We have developed a novel analytic approach based on a mix of Bayesian statistics and our own proprietary media tools, which help us to predict and optimize the offline and online media. Key features include Monte Carlo scenario analysis, forecasting under uncertainty (constrained media, lack of attribution data etc.), quantifying short and long-term channel effects. Using a client case study, we will highlight our approach and a framework to make timely and effective business decisions.
Speaker: Chris Dickey, SVP, Director of Analytics & CRM, The Martin Agency, David Henkel, Manager - Digital Optimization, Penske, Neeraj Kulkarni, Senior Statistician, Martin Agency, John Busbice, Founder, MIDA

[ Top of this page ] [ Agenda overview ]

3:50-4:10pm

All level tracks Track 1: Predictive Project Management
Case Study: Tangent Design Engineering
Applying Predictive Analytics to Improve Project Management

Project managers today are struggling to keep control over multiple projects in today's complex work environment. Almost all projects are in a state of constant change and managers are tasked with making continual adjustments to reflect competing priorities, scheduling changes, and resource allocation. It's estimated that even small teams will require billions of calculations each month to accurately adjust their project schedules. Businesses today need predictive project management solutions that use advanced statistical algorithms to automatically adjust and accurately predict project schedules to positively impact the bottom line. Tangent Design Engineering is a great example.
Speaker: Brett Bender, Principal Software Engineer, LiquidPlanner
3:50-4:10pm

Track 2: Self Updating Models
Case Study: Ace Cash Express
Data Driven Modeling

Models break down over a period - we call this stability of the model. The more the modeler gets creative, the faster the model breaks-down. In this presentation, you will learn:

The techniques and the benefits of building a self-updating model, so you never have to worry about models collecting dust
Ways to regulate an automated model, so human control is not lost
Methods you can employ in reporting against ever-changing model

Speaker: Senthil Ramanath, Head of Credit Risk Analytics, Ace Cash Express

3:50-4:10pm

All level tracks Track 1: Marketing Lab
Case Study: Predicting Sales through Brand Research Data
Closing the Chasm Between Marketing and Sales

For marketing managers, consumer insight is what helps build brands. The brand's Key Performance Indicators (KPIs) are the measures that they live by. Every brand deploys extensive tracking of these brand performance measures and consumer sentiment across categories and markets at regular intervals.

However, in today's challenging economic environment, spends on research are being put under the scanner, and the ROI on such initiatives is often questioned. Increasingly, marketing managers are being challenged by their sales and operations planning counterparts on the efficacy of such measures and their relevance to sales. Marketing Mix Modeling and other advertising testing models explain the effects of promotions on sales, but they don't establish the connect with the brand's KPIs. This puts added pressure on marketers with their decision to conduct research for what in effect is seen as a status check on brands with no direct implications on business.

Through our case study on a global beverage major, we will showcase how consumer tracking research can predict total brand sales potential and provide foresight to the sales organization. We will be taking this initiative on behalf of the marketing fraternity.
Speaker: Amitabh Bose, Sr. Director – Capability, WNS Global Services

[ Top of this page ] [ Agenda overview ]
4:10-4:35pm

Break / Exhibits

[ Top of this page ] [ Agenda overview ]
4:35-5:20pm

All level tracks Track 1: PA for Financial Regulations
Case Study: Goldman Sachs
Utilizing Predictive Analytics to Analyze Changing Regulation

As regulatory expectations increase, organizations must utilize predictive analytics techniques to predict the impact of regulatory change. Regulations such as the Volker Rule, Anti-Money Laundering and Model control regulation require companies to more effectively conduct surveillance and analysis. This session focuses on:

Learning how to effectively analyze and predict the impact of new regulation to your business
Take away practical tips to generate accurate alerts, model data feeds against your products and tune thresholds
Determine the right technology to resolve common data challenges and maintain clean data
Learn how to avoid mishandling complex analyses and reporting

Speaker: Vikas Agrawal, Global Head of Analytics, Goldman Sachs

4:35-5:20pm
Track 2: Healthcare Analytics
Case Study: Pfizer
Right Medicine, Right Patient

Can predictive modeling improve patient care? A wealth of data exists in large healthcare databases on patient disease characteristics and their response to specific treatments. Max will discuss some of the technical and non-technical issues in providing care providers with quantitative results related to how individual patients might response to therapies.

Speaker: Max Kuhn, Director of Nonclinical Statistics, Pfizer

[ Top of this page ] [ Agenda overview ]

5:25-6:10pm
All level tracks Track 1: Economic Research and Forecasting
Case Study: Rebellion Research
Applying Machine Learning to Global Economics

Global Economic research is provided by the biggest banks and ratings institutions, yet the predictive powers of these firms is notably and infamously low. Bloomberg reported that major Wall Street research is typically 46% accurate. However, this is a field that deals with an extremely large and accessible data set. One that is ideal for machine learning. Yet, there are no machine learning powered research or ratings firms on Wall Street. Robotic ratings and research firms could dominate this field, yet are not even in existence today.

Speaker: Alexander Fleiss, Chairman, Rebellion Research
5:25-6:10pm
Track 2: Clinical Healthcare
Case Study: Sisters of Mercy Health Systems
Framework for Detection of Clinical States & Disease Onset Using Electronic Health Record (EHR) Data

This case study describes the application of predictive analytics to the detection of disease onset and clinical states through the use of electronic health records (EHR). The framework presented here aims to improve prediction of a patient's risk for developing severe sepsis and septic shock through a risk score generated as a function of measurements of patient vitals over time. A risk score threshold of 0.71 was found to yield the highest sensitivity while minimizing false negatives in the patient database. This predictive model can also be generalized to predict outcomes of other application domains.
Speakers: Jeni Fan, Associate, Booz Allen Hamilton & Juergen A. Klenk, PhD, Principal, Booz Allen Hamilton

6:10-7:30pm

Reception / Exhibits

7:30-10:00pm

Boston Predictive Analytics MeetUp
Lightning Talks: "Data Deluge!, U.S. Jobs Outlook, D3 DataViz, R Data Mining, Random Forests"

The goal of the Meetup group is to help the local community further it's understanding and proficiency regarding Predictive Analytics through informative lectures, hands-on tutorials, and networking events. Our group has three main focal points: business applications, advanced mathematics, and computer science. Past events have included sentiment analysis, web content recommendations, social media and network analysis; as well as several events pertaining to the Big Data / Hadoop ecosystem.

Boston's Meetup Community: John Verostek
Tapping the Data Deluge!: Jeffrey Breen
U.S. Jobs Outlook: John Muller
Data Visualizations using D3: Lynn Cherny
Data Mining with R / Rattle: David Weisman
Random Forests Case Study: Dan Gerlanc

www.meetup.com/Boston-Predictive-Analytics

[ Top of this page ] [ Agenda overview ]

Conference Day 2: Tuesday, October 2, 2012

8:00-9:00am

Registration & Breakfast

[ Top of this page ] [ Agenda overview ]

9:00-9:45am

Keynote
Putting IBM Watson to Work

IBM's Watson captured the imagination of over 34 million viewers when it beat the all time champions of the US game show, Jeopardy!. To do so, it navigated the complexities of human speech, churned through 200 million pages of unstructured data in under 3 seconds, delivered a confidence based response, all while learning and getting smarter with each outcome. But as impressive as this accomplishment was, it was only the beginning. IBM is working with leading organizations across industries to put Watson to work. The possibilities are endless! Join Bob Jewell, Director of Business Development and Partnerships for IBM Watson Solutions, in an engaging discussion of how IBM Watson can fundamentally transform the way businesses and individuals make decisions and how next generation systems will be designed.
Speaker: Robert Jewell, Director, Global Business Development & Partnerships, IBM Watson Solutions

[ Top of this page ] [ Agenda overview ]

Deloitte 9:45-10:05am

Platinum Sponsor Presentation
Forensic Analytics – Discover insights that can help you move forward

Internal audit, compliance groups and the like are increasingly being asked to do more work with less resources. In the world of forensic analytics we are utilizing advanced analytic techniques to help those charged with the tasks of investigating fraud, waste, abuse and corruption combat their resource constraints. Forensic Analytics applies a variety of techniques and methodologies to transform disparate data sources into forensic insights for timely action. But what happens when the fraud or corruption has already occurred? Utilizing predictive modeling to gain forensic insights into the vast amount of data can yield promising results, especially when looking at damages. This presentation explores the use of predictive modeling to look at what could have been, had unfair lending practices at a large financial institution not occurred.
Speaker: Carol Tannous, Senior Manager in the Data Analytics practice, Deloitte Financial Advisory Services LLP.

[ Top of this page ] [ Agenda overview ]

10:05-10:40am

Breaks / Exhibits

[ Top of this page ] [ Agenda overview ]

10:40-11:25am

Expert Panel
Big Data for Predictive Analytics

If Big Data begs the question, "What to do with all this data?" predictive analytics answers, "Learn from it to predict behavior." But just how much predictive payoff comes with going so big? This expert panel will address the new demands on predictive analytics solutions and best practices as data grows to enormity, and will recommend tactics to fully leverage data's growing magnitude to improve the business performance of predictive analytics initiatives.
Moderator: Eric Siegel, Ph.D., Conference Chair, Predictive Analytics World
Expert Panelists: Satish Lalchand, Director, Deloitte Financial Advisory Services LLP
Jason Verlen, Director, SPSS Product Strategy & Management, IBM Software Group
Eric Feinberg, Senior Director of Mobile, Media and Entertainment, ForeSee

[ Top of this page ] [ Agenda overview ]

Mu Sigma
11:30-11:40am

Gold Sponsor Presentation
Data – Hurdle or Springboard? How to leverage data to transform your business?

Organizations look at data as a requirement to solve business problems on hand and often consider lack of data as a hurdle. Their data strategy is primarily driven by the business problem that they need to solve. (Problem driven approach: Business problem -> Hypothesis formation -> Data collection -> Hypothesis testing). The lack of data to answer key business questions, along with advances in technology have always served as a springboard for new data e.g., RFID, telematics, gene sequencing etc. Recently, analytically savvy organizations have discovered that new data sources could be used in more innovative ways than to just solve the problem in hand. This has led to a new paradigm in data strategy where the data story starts with the data in hand and not from the business problem (Discovery driven approach: Agenda less observations -> Pattern anticipation -> Data identification and collection -> Hypothesis testing -> Business Opportunity). The presentation focuses on how analytically savvy organizations incorporate both paradigms to leverage data in innovative ways to create game changing opportunities.
Speakers: Mukund Raghunath, Geography Head, Mu Sigma Inc.

[ Top of this page ] [ Agenda overview ]

IBM 11:45am-12:30pm

All level tracks Sponsored Lab
Lab Session: Live Topical Demo
Building Customer Relationships Through IBM SPSS Predictive Analytics

Building life-long customer relationships is a key goal for ING U.S. Financial Services. To build stronger customer relationships, we recently launched a new customer experience program across our U.S. contact centers. At the heart of this program is a predictive analytics capability that is designed to maximize the value of customer touchpoints. In the first 12 months since launch, predictive analytics has guided more than 300,000 customer interactions and resulted in 24,000 positive customer actions. This presentation will describe how predictive analytics has been used to deliver the right message to the right customer at the right time.
Speakers: Jason Verlen, Director, SPSS Product Strategy & Management, IBM Software Group
Tom Hamilton, Director, Business Intelligence Competency Center, ING U.S.

[ Top of this page ] [ Agenda overview ]

12:30-1:30pm

Lunch / Exhibits

1:30-2:15pm

Special Plenary Session
Becoming an Ace with a Robot as your Wingman

Humans and computers have strengths that are more complementary than alike – to the point where a sophisticated algorithm may be the best "2nd person" to put on a complex task. Yet, our and computer analytic weaknesses are surprisingly severe. To explore how to improve the man/machine partnership, we compare and contrast natural and artificial intelligence, with special attention to the growing realization of how challenging it is to think truly rationally.

Speaker: John Elder, CEO & Founder, Elder Research, Inc.

[ Top of this page ] [ Agenda overview ]

WNS 2:20-2:25pm

Gold Sponsor Presentation
"Quest for Big Insight"- Path to Actionable Insights from the Data Deluge

All firms today seek to leverage all of their data to make fact-based actionable business decisions. But how many are truly getting the ROI they want or the insights they need?

Technology has made it possible to harness a wealth of transactional data and even as industry was coming to terms it, the gamut of Big Data has opened up a virtual deluge, pun intended.

Insights however continue to remain the key to creating competitive advantage; in this session, we discuss how companies have leveraged WNS' proprietary analytics decision engine WADE to derive insight from data and drive customized and scalable business solutions enabling fact-based choices.

Speaker: Sanjit Bhoumick, Senior Vice President – Sales, WNS Global Services

[ Top of this page ] [ Agenda overview ]

2:30-2:50pm

All level tracks Track 1: Sales Force Optimization
Case Study: Hewlett-Packard
Sales Productivity Analysis - Optimizing Time Spent to attain Maximum Returns for Sales Reps

Any organization needs a very efficient and effective sales force in order to be competitive. Among various efficient and effective parameters, we focused on time spent pattern of the sales representatives and how it would influence their sales performance (Win Rates). To achieve this, a non-linear causality between the Win Rates and Time Spent pattern was designed. The analysis predicted maximum win rates that can be achieved with optimal time spent. It helped to recommend future strategies for sales organization for better reallocation of time spent across job activities, which in turn would help sales to enhance ROI.
Speaker: Guruprasad Srinivasan, Project Manager, Hewlett-Packard & Subhamitra Chatterjee, Marketing Analyst, Hewlett-Packard
2:55-3:10pm

All level tracks Track 1: Predicting B-2-B Buying
Case Study: Hewlett-Packard
Behavioral Analysis of Account Purchases and their Predictability

Analyzing purchase behavior in the enterprise space is a nascent area of research - to determine what a customer account is likely to buy next, when it will buy, what is the probability of winning a deal and finally, the expected value of the purchase. Our holistic solution dynamically captures changes in influencing variables in real time, acting as an early warning system that helps sales teams prioritize opportunities, take necessary action and allocate resources optimally. This has assisted the sales organization at HP to pursue the most profitable opportunities within the right accounts, maximize profitability and improve market share.
Speaker: Govindarajan Krishnaswamy, Analytics & Solutions, Aparna Seshadri, Business Analytics Solution Development, & Ronobijay Bhaumik, Lead, Sales Processes and Execution Portfolio, Global Business Intelligence, Hewlett-Packard
2:30-3:15pm

Track 2: True Lift Modeling
Case Study: Staples
True-Lift Modeling: Mining for the Most Truly Responsive Customers and Prospects

Stop spending direct marketing dollars on customers who would purchase anyway!

True-lift modeling can identify:

which customers will purchase without receiving a marketing contact
which customers need a direct marketing nudge to make a purchase
which customers have a negative reaction to marketing (and purchase less if contacted)

This discussion will describe:

the pros and cons of various approaches to true-lift modeling
metrics for evaluating mdoel performance
the basic requirements needed to succeed with true-lift modeling
scenarios where this modeling method is most applicable

See the white paper on this topic

Speaker: Jane Zheng, Chief Scientist, Focus Optimal

[ Top of this page ] [ Agenda overview ]
3:15-3:55pm

Breaks / Exhibits

[ Top of this page ] [ Agenda overview ]

3:55-4:15pm

All level tracks Track 1: Financial Services
Case Study: Commerzbank, Lloyds, Ursus Advisors
Optimization in Financial Services

Financial institutions are improving business performance by using optimization techniques in additional to behavioral and statistical modeling. Hear how retail banks are using prescriptive along with predictive analytics to get more customers, get them more profitably, price them appropriately, and retain them strategically.

We will cover the barriers to adoption and introduce an organizational readiness checklist. Attendees should leave ready to kick off the projects internally to implement optimization.

Speaker: Frank Bria, Research Director, Tower Group

4:20-4:40pm
All level tracks Track 1: Vendor Recommendations (Beyond Product Recs)
Case Study: Intuit and Mint
Restaurant Recommendation using Financial Likeness

Imagine you are in a new neighborhood and want to discover the 'right' restaurant for 'you'. Instead of reading a host of reviews, we present a recommender system that learns from your historical spend transactions. The system predicts appropriate restaurants from financial transactions of users in Mint, a personal finance tool from Intuit. In contrast to domain specific apps, our recommender can be scaled across different verticals which have financial transaction data from users and hence is more general purpose. We will describe the challenges in using spend data for recommendation, scaling to millions of merchants and transactions with Hadoop and Mahout, and evaluating the performance of the app.

Speaker: Saikat Mukherjee, Data Scientist, Intuit
3:55-4:40pm

Track 2: Detecting Cannibalization
Case Study: TripAdvisor
Cannibalization Analysis Using Matched Pairs at TripAdvisor

In 2010, TripAdvisor launched our Business Listings product that sends travelers directly to hotel web sites, bypassing on-line travel agencies. Is the new product stealing revenue from our traditional CPC business? This case study explains why the question is hard to answer, describes the analytic approach we employed, and reveals the answer to the question.

Speaker: Michael Berry, Business Intelligence Director, TripAdvisor

[ Top of this page ] [ Agenda overview ]
4:45-5:30pm

All level tracks Track 1: Crowdsourcing Data Mining
Machines Learn, But Can They Teach?

In 2012, a British particle physicist, a data analyst for the National Weather Service in Washington, D.C., and a graduate student from Germany (none of whom had a background in education) collaborated on the Kaggle platform to win the $100,000 Hewlett Foundation Automated Student Assessment Prize, a competition to build an innovative algorithm that could score students' essays used in state standardized tests. Software scoring programs do not independently assess the merits of an essay; instead they predict, very accurately, how a person would have scored the essay. This is a critical distinction because it means that the software replicates the same scores as trained educators for significantly less time and money. The winning algorithm outperformed the current state-of-the-art in commercial grading software and achieved the same level of agreement with a trained human grader as two human graders have with each other. Anthony Goldbloom, the founder and CEO of Kaggle, discusses what these results mean for the future of education technology and crowd-sourced competitive analytics.

Speaker: Anthony Goldbloom, Founder & CEO, Kaggle
4:45-5:30pm

Track 2: Customer Valuation
Case Study: United Group Holdings
Value Proposition Segmentation (VPS) Method

In today's competitive market, effective management of customer relations lies in the ability to optimize the dual creation of firm (shareholder) and customer value. Accordingly, the challenge for many companies is to be able to understand customers by their needs to deliver the winning value proposition profitably. This session will show how our proposed VPS model addresses the basic managerial concern of balancing relationships from both the seller's (customer loyalty) and the buyer's (customer benefit), by considering both the service provider's financial performance (i.e. customer value to the firm) and the value customers receive from the the provider's offerings.

Speaker: Amjad Zaim, Chief Executive Officer, Cognitro Analytics

[ Top of this page ] [ Agenda overview ]
Post-Conference Workshops: Tuesday, October 2, 2012

Three hour evening Workshop – 6:30-9:30pm
R Bootcamp: For Newcomers to R

Click here for a detailed workshop description

Instructor: Max Kuhn, Director, Nonclinical Statistics, Pfizer
Post-Conference Workshops: Wednesday, October 3, 2012

Full-day Workshop
Modeling Methods: The Best & the Worst of Predictive Analytics: Predictive Modeling Methods & Common Data Mining Mistakes

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

24 сен 2012 20:32
Хмур
Модератор
Модератор
Сообщений: 3531
Больше
Обзор конференций по анализу изображений и распознаванию образов #22229
Слава>> ..use of predictive analytics has come a long way in recent years

Не могли бы вы рассказать, в чем именно был прогресс, если это ваша профессиональная область. наверное имеются в виду
и области приложения и техники.

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

24 сен 2012 20:50 24 сен 2012 20:59 от slava.
slava
Старожил
Старожил
Сообщений: 1205
Больше
Обзор конференций по анализу изображений и распознаванию образов #22231

Хмур пишет: Слава>> ..use of predictive analytics has come a long way in recent years

Не могли бы вы рассказать, в чем именно был прогресс, если это ваша профессиональная область. наверное имеются в виду
и области приложения и техники.


Это же - не мой доклад
А предсказания это - часть усиления в случае процессов
Мы используем наши методы обучения распознаванию для построения неявных моделей процессов и вскрытия закономерностей динамики
А среди этих аннотаций что-то мельком видел про чел-маш комплексы.
На первый взгляд показалось, что это может быть близко. Но деталей нет, а в них, как всегда, главное
Да и с остальным есть что-то любопытное - в общем, люди не спят

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

24 сен 2012 22:05 24 сен 2012 22:06 от Хмур.
Хмур
Модератор
Модератор
Сообщений: 3531
Больше
Обзор конференций по анализу изображений и распознаванию образов #22234
спасибо, но я ожидал названия конкретных методов, методик, алгоритмов, теорий (хотя бы как ключи),
в смысле как именно не спят люди, а что они не спят и что в деталях главное (а именно про них, напомню,
и спрашивалось), и что, даже, что предсказание суть интеллектуальная функция, - так оно вроде не может
не быть известным всем.. хороший прагматик - это содержательный прагматик..

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

25 сен 2012 08:06 25 сен 2012 09:05 от slava.
slava
Старожил
Старожил
Сообщений: 1205
Больше
Обзор конференций по анализу изображений и распознаванию образов #22244

Хмур пишет: спасибо, но я ожидал названия конкретных методов, методик, алгоритмов, теорий (хотя бы как ключи),
в смысле как именно не спят люди, а что они не спят и что в деталях главное (а именно про них, напомню,
и спрашивалось), и что, даже, что предсказание суть интеллектуальная функция, - так оно вроде не может
не быть известным всем.. хороший прагматик - это содержательный прагматик..


У нас с вами разные представления об интересном
Мне интересно интересное, а не многознание - у меня с детства проблемы с памятью
Что касается прагматики, то, возможно, я неправильно использую этот термин.
Один весьма неглупый, по моему мнению, человек однажды назвал меня абсолютным прагматиком. Мне это понравилось. Может быть, это была шутка, но я не стал разбираться
Предсказание - конечно, интеллектуальная функция, одна из базовых, меня интересует. Но - не просто так, а виде эффективной модели ее в первую очередь. Об этом как-то мало говорят. Конечно, если что-то попадается или попадется, я это фиксирую, но пока больше разочарований. Пример - все та же биржевая игра. Впрочем, вы про это, вроде, неплохо знаете

Пожалуйста Войти или Регистрация, чтобы присоединиться к беседе.

Модераторы: Inex
Время создания страницы: 0.696 секунд
Работает на Kunena форум