We introduce a novel model-agnostic system that explains the behavior of complex models with high-precision rules called anchors, representing local, "sufficient" conditions for predictions. We propose an algorithm to efficiently compute these explanations for any black-box model with high-probability guarantees. We demonstrate the flexibility of anchors by explaining a myriad of different models for different domains and tasks. In a user study, we show that anchors enable users to predict how a model would behave on unseen instances with less effort and higher precision, as compared to existing linear explanations or no explanations.

In machine learning, rule models are one of the most popular choices when model interpretability is the primary concern. Ordinary, a single model is obtained by solving an optimization problem, and the resulting model is interpreted as the one that best explains the data. In this study, instead of finding a single rule model, we propose algorithms for enumerating multiple rule models. Model enumeration is useful in practice when (i) users want to choose a model that is particularly suited to their task knowledge, or (ii) users want to obtain several possible mechanisms that could be underlying the data to use as hypotheses for further scientific studies. To this end, we propose two enumeration algorithms: an approximate algorithm and an exact algorithm. We prove that these algorithms can enumerate models in a descending order of their objective function values approximately and exactly. We then confirm our theoretical results through experiments on real-world data. We also show that, by using the proposed enumeration algorithms, we can find several different models of almost equal quality.

All things change, and you must adapt over time. Ongoing monitoring of machine learning fraud detection systems is imperative for success. As populations and the underlying data shift, expected system inputs degrade and therefore have an impact on overall performance. This isn't unique to machine learning systems; rule-based systems have the same challenge. But newer machine learning methods can adapt to new and unidentified patterns as underlying changes occur. This eliminates some, but not all, of the machine learning retraining and evaluation steps.

Stephen Hawking said, "The development of full artificial intelligence could spell the end of the human race….It would take off on its own, and re-design itself at an ever increasing rate. Humans, who are limited by slow biological evolution, couldn't compete, and would be superseded." Ever since its genesis, there have been conflicting views and concerns on the potential enhancement or doom that it can cause to human civilization. While some experts believe that this technology will advance and augment our intelligence, some like Bill Gates have expressed concerns on how a machine's intelligence becomes strong enough to be a concern. For now, let's take a look at the current trends of AI and where it is headed to in 2018: If 2017 was the year where the warnings from Elon Musk and Stephen Hawking about the potential evil from AI clashed with predictions from Mark Zuckerberg and Bill Gates on its potential good, 2018 will be the year when the debate shifts to its practical utility.

When you search for security data science on the internet, it's difficult to find resources with crisp and clear information about the use cases, methods and limitations in Information Security (hereby referred to as InfoSec). There's usually always some marketing material attached to it. So, I thought of summarising my knowledge and InfoSec experience in this article. When the attackers are within the enterprise network, they first need to figure out where they are. Once they accomplish this, they move towards their targets, and carry out the attack.

If you want to see data being put hard at work, look no further than artificial intelligence and machine learning – never before has timely, accurate data been so critical. A recent survey released by Narrative Science finds 61 percent of enterprises have already implemented AI within their businesses. At least 71 percent said AI was part of their enterprises' innovation strategies. At the same time, 90 percent of respondents in the business intelligence function reported that they would be interested in incorporating AI to make their data and analytics tools smarter. At the same time, AI and machine learning are not overnight projects in themselves – they take considerable time and expertise to get rolling, and constantly need refreshing and updating.

Machine learning has had fruitful applications in finance well before the advent of mobile banking apps, proficient chat bots, or search engines. Given high volume, accurate historical records, and quantitative nature of the finance world, few industries are better suited for artificial intelligence. There are more uses cases of machine learning in finance than ever before, a trend perpetuated by more accessible computing power and more accessible machine learning tools (such as Google's Tensorflow). Today, machine learning has come to play an integral role in many phases of the financial ecosystem, from approving loans, to managing assets, to assessing risks. Yet, few technically-savvy professionals have an accurate view of just how many ways machine learning finds its way into their daily financial lives.

System detects patterns in Email About likely markers of spam Detected Pattern Emails with "Earn Cash" More likely to be spam email Can use such detected patterns to make automated decisions about future emails Example: Email Spam Filter "Earn Cash" "Earn Cash" detected in 10% of Spam emails 0% of wanted emails Identification Improves Algorithm improves in performance In auto-identifying spam As it is able to examine more data And find additional indicia of spam Algorithm is "learning" over time from additional examples Example: Email Spam Filter "Free" Probability of Spam Contains "Free" 70% Spam Contains "Earn Cash" 90% Spam From Belarus 85% Spam For some (not all) complex tasks Requiring intelligence Intelligent Results Without Intelligence Can get "intelligent" automated results without intelligence By finding suitable Proxies or Patterns People use advanced cognitive skills to translate Proxies for Intelligent Results Without Intelligence Google finds statistical correlations by analyzing previously translated documents Statistical Machine Translation Produces automated translations using statistical likelihood as a "proxy" for underlying meaning Detecting Patterns Proxy Principle for Automation That can serve as Proxies For some underlying Cognitive Task Learning Machine Learning Main Points Pattern Detection Data Self-Programming Summary Major AI Approaches Two Major AI Techniques • Logic and Rules-Based Approach • Machine Learning (Pattern-Based Approach) Hybrid Systems • Many successful AI systems are hybrids of • Machine learning & Rules-Based Hybrids • e.g. Self-driving cars employ both approaches • Human intelligence AI Hybrids • Also, many successful AI systems work best when • They work with human intelligence • AI systems supply information for humans Humans Computers Technology Enhancing (Not Replacing) Humans Humans Alone Computers Alone Examples of AI in Law Today • Machine Learning • AI in Litigation - E-Discovery and "Predictive Coding" • Natural Language Processing (NLP) of Legal Documents • Automated contract analysis • Predictive Analytics for Litigation • Machine Learning Assisted Legal Research • Logic and Rules-Based Approaches • Compliance Engines • Expert Systems • Attorney Workflow Rule Systems • Automated Document Assembly Limits on Artificial Intelligence • Artificial Intelligence Accomplishments • Automate many things that couldn't do before • Limits • Many things still beyond the realm of AI • No thinking computers • No Abstract Reasoning • Often AI systems Have Accuracy Limits • Many things difficult to capture in data • Sometimes Hard to interpret Systems Questions Harry Surden Associate Professor of Law University of Colorado Law School Affiliated Faculty, Stanford CodeX Center Twitter: @HarrySurden Email: hsurden@colorado.edu

For roughly a decade, the SEO industry has been consistently moving further away from discussions about the internal workings of the search engines, starting around the time of the first Panda update. This decision to focus on more strategic and long-term SEO approaches has been a positive one, overall. Marketing strategies built entirely on the whims of a search engine aren't built to stand the test of time. However, there's no denying the value that came from understanding and predicting how search engines would react to certain kinds of tweaks to the HTML or to the backlink profile in the early days of SEO. Most of us have assumed that those days ended when Google embraced machine learning.

However, this reasoning is based on the assumption that once a machine learning algorithm has been developed to solve one problem, that same algorithm can be easily applied to solve a different problem.