What did the buffalo say to his son when he left for college?

“Buffalo buffalo Buffalo buffalo buffalo buffalo Buffalo buffalo.” Prof. Pushpak Bhattachararya kick-started our NLP course at IIT Bombay with the ‘buffalo’ sentence. It worked perfectly for him; he could witness our amusement when he declared that it is a grammatically correct statement! Our ways to communicate continue to amuse me at times. As a matter of fact, it is often debated that language played the most important role in human evolution. In this article, I argue that language will also play the pivotal role in evolution of artificial intelligence.
A criteria to determine intelligence? Understanding a language. I am not talking about understanding grammar; that is a simpler task. Grammar is what we collectively, as a society, decide. Language is characteristic not of a society, but of a person. Each individual has a unique way to use and misuse language. That doesn’t form a communication barrier for us, though. Our ability to understand each other is remarkable. Can a machine be trained to have a similar ability?
As I mentioned in my last article, Machine learning has become a synecdoche for Data science; our so-called intelligent machines are limited to making direct inferences from data. Recently, iOS was blamed to be sexist, because when you type-in leadership roles like ‘CEO’, ‘CFO’ or ‘CTO’ using default keyboard, the emoji it suggests will be the default male in a suit. Did it choose to be sexist? It probably just reflects the data from conversations of millions of humans; it doesn’t choose to be good or bad, it only chooses to be like humans. Intelligence is more than this.
Arnold, in HBO’s Westworld, has the notion of a ‘consciousness pyramid’:

These layers of the pyramid portray the depth of consciousness. It made me wonder, does intelligence have similar depth to it?
Currently, we are working on harmonizing Point-Of-Sales (POS) data. Let me introduce you to my favorite member of our team, we call him Frank. Frank is 2 months old now. He is extremely good at understanding semi-structured sequences of words (To call our inputs as sentences would be incorrect). Why do we call him Frank?
“It’s alive. It’s alive… It’s alive,” – Frankenstein
What makes Frank lively? It is not the tendency to make the same decisions which humans make. We couldn’t settle for that, because we didn’t have sufficient data for supervised learning, and the data which we had has quite a few errors. Harmonization of POS data is simply too difficult and time consuming of a process to be done manually. Frank was initially designed as an unsupervised learner. It started by learning a vocabulary which we created using corpuses like OpenFoodFacts, DBPedia, WordNet, etc. Frank structures its vocabulary as word clusters by guessing the semantic distance of words using word embedding. The second level of learning comes into the picture when a new client is on-boarded; Frank learns the domain of the client depending upon how the client wants its data to be harmonized. Essentially, Frank biases its vocabulary to assign weightages to words as per their importance to the client.
At this stage, we started getting decent results, so we wondered, can we add supervision to it? After few failed attempts using mainstream techniques, we ended up building a layer which mimics survival instinct.
Philosophy of this third layer: “If it saves your life, 2+2=5”. We spawn multiple supervised learners at this level, each one of them aims at surviving consistently in at least one domain. Each of these classifiers behaves like the human notion of ‘stereotypes’ for Frank; each classifier knows how to behave in a particular context. To be precise, each of these classifiers is an over-fit. These classifier have a role similar to weak classifiers in boosting algorithms; difference being that weak classifiers are functionally weak, Frank’s classifiers are partially-trained directional fits. The benefit of using a functionally complex classifier as a weak classifier is that over time and data, it will stabilize to a less weak state, which is not the case with a conventional weak classifier.
How does Frank know which stereotype should be used when? The third level represents the confusion in Frank’s head, and the resolution comes at the fourth level; Frank has to make the ‘choice’. Frank adapts over time to understand which stereotype works better in which scenario. This latter half of Frank is a greedier and cleverer extension of “Random Forests” classifier. Depending on the domain, client, and the quality of feedback given by people who manually verify Frank’s results, classifiers might either be chosen over others, decommissioned, or put in cryosleep until they get better results. Since individual classifiers capture local patterns perfectly, we have simulated Long Short-Term Memory (LSTMs) without explicitly using the construct; Frank remembers local patterns, but the over-fit involved in finding those patterns is balanced with the choice Frank faces in selecting those.

Frank is ready to do his job of harmonization. Is Frank ready to face the world? Not yet. He can do his specific job very well, but the human world and our language has too much creativity for Frank to understand. Frank can understand structure and context, but it is not trained to understand metaphors, similes, sarcasm, etc. Yet, I feel Frank has made an important step in the direction which future generation of intelligent machines will be following.
Let’s end the article hoping that someday Frank will laugh at my geeky buffalo joke:
What did the buffalo say to his son when he left for college? Bison.

Automation, Predictive Analysis and AI will Transform Life Insurance in The Digital-Age

Do you remember the movie Elysium? Yes, the gigantic space habitat located in Earth’s orbit, which had the medical machine—Med-Bays that cured all diseases, even reversed the aging process by collecting the right data from blood samples. Future has always been a fountain of creative possibilities, for filmmakers and storytellers. Today, their visions have inspired many businesses, like wearable tech; connected devices are a reality. Wearable biometric sensors, such as FitBit can track information related to health and fitness.
As new technologies emerge, consumer trends and preferences change. Similarly, for the traditionally cautious and heavily regulated insurance industry, data technology is transforming their core business model. With the rise of Artificial Intelligence and Machine learning, insurance activities are becoming more automatable. Hence, the biggest opportunity for traditional insures will be to effectively capture and analyze data to better manage financial risks, develop new pricing models and mitigate personal health dangers.
Automation and Predictive Analysis will motivate people to improve their lifestyles, by gathering data from sensors to analyze people’s exercise habits, and vital signs, including their heart rate blood pressure, body temperature and an array of other health metrics. Through rule based algorithms, the accumulated data helps generate information that can recommend behavioral improvements (diet needs, or will suggest for increased activities in daily life)—predictive and prescriptive data and evaluate the results of these recommendations
Artificial Intelligence in Insurance:
The role of AI in logical intelligence is well documented and already being implemented in many industries. What has not been discussed in depth yet is emotional intelligence in relation to AI. This will be an even stronger future driver in the success of an automation/analytics solution for the insurance industry. How? Emotional intelligence can be initially derived from non-verbal gestures, including facial expressions or different tones of voice. Rather than relying on what an individual inputs as their activity (if the activity cannot be recorded automatically through hardware tracking), AI can possibly leverage EQ to ‘see through’ manual inputs of activity to accurately predict the likelihood of the individual successfully accomplishing their tasks. Though research in this area is still quite early, it is promising that insurance organizations are looking towards this direction. It will not only lead to a more efficient insurance industry as a whole but also a much more accurate view into an individual’s health status.
The challenges remain, but we are moving swiftly into a future in the insurance industry and other industries as well where quantifying human health will change the way many organizations operate, especially their interaction with consumers.
 
 
 

5 Practical Ways to Embrace Automation in your organization

Almost all coffee making recipes require the beans be ground, and then mixed with hot water long enough to allow the flavor to emerge, but not so long as to draw out bitter compounds.
And then came coffee pots, vacuum pots, moka pots, electric coffee maker, and what not! Its purpose? To infuse until the desired strength brew was achieved. Similarly, any organization that wishes to achieve the perfect blend of growth and sustainability must embrace automation.
While the benefits of enterprise automation initiatives have been well documented, in order to successfully reap value, organizations must understand that it will take more than just implementing tools. This is because the secret recipe is more strategic, and not tactical. Here are five strategic initiatives organizations must undertake to successfully introduce automation:

Educate

Organizations must understand the “can do & cannot do’s” of automation, before embracing it. Quite often, unrealistic expectations on technology kills future applications. It is therefore important to setup internal structures to educate, and evangelize the idea of automation. Also, automation is here to stay. It’s the key to our utopian, neon light future. So, the misconceptions about automation killing jobs must be addressed to ensure success.

Structure

With unlimited potential on offer, organizations would be well served to setup a Centre of Excellence to oversee the successful application of automation technologies. With enterprise software vendors racing to enrich their software suite’s capabilities with embedded AI, license costs in large disparate landscapes are set to be impacted. Instead of bearing the expensive license costs YoY, it would be better to implement fit-for-use automation use-cases over existing investments.

Functional Focus

While centralizing investments and developments, it is also important to prioritize investments into specific functions. Some functions like Procurement, Finance and IT are excellent starting points. While automation platform vendors are bundling use-cases in these functions, specific focus must be laid on leveraging automation for continuous improvement. This is the key to leverage its full potential.

User Experience

The need for positioning the user in the center of all automation initiatives is key to success. AI based systems require significant training to deliver value to users. It is therefore important to provide a superior user experience to drive engagement.

Choosing the right Partner

Given what’s at stake, it is recommended not to go with an unproven firm. In this journey, your preferred partner should display an eagerness to meet your organization’s challenges, and bring in rich domain, technology, industry and expertise. Some vendors deliver platform expertise without an industry focus, while some niche industry focused solutions. The key to success is to bring in cross functional experts to build fit-for-use solutions.
Final thoughts: Humans and machines can be partners and collaborate for shaping the future. As Bill Gates pointed out, “The first rule of any technology used in a business is that automation applied to an efficient operation will magnify the efficiency. The second is that automation applied to an inefficient operation will magnify the inefficiency.”

3 Data Driven Challenges in Consumer Packaged Goods Companies

As one of the biggest drivers of change and innovation, the Consumer Packaged Goods (CPG) companies have always launched innovative products to meet a bolstering array of human wants. This rapid growth was made achievable and profitable only because of continuous product innovation. However, the past is no guide to the future.
According to Nielsen (NLSN) report, 2016, “only 15 percent of newly introduced consumer packaged goods succeed in the market.” Because the consumers are increasing finding the digital medium— e-commerce, Omni-channel retailing, and mobile platforms more compelling than the traditional stores. And, with this eruption there is a need for intelligent innovation in CPG industry.
Also, with this transition comes the enormous amount of ‘Big Data’, which can be used to drive business decisions, and can also help create a compelling and integrated view of the consumer experience. But doing so is not as easy it seems. Before turning any type of data into valuable insights, the CPG industry needs to address the following barriers:

Getting the right data

CPG companies today have abundant data, sourced internally and externally. But this doesn’t guarantee a great value to enterprises’ sales information. Since data obtained for entities may be incomplete or can be in an incorrect format, the quality of data is questionable.
So, when enterprises crawl in to gain a timely visibility into their sales information, it leads them to faulty conclusion. Data-driven decisions are only as good as the quality of data you collect. Using data to drive business decisions requires the right data collection system that can transform these data into actionable insights.

Gaining the proficiency to use data

Having more data doesn’t necessarily lead to actionable insights, and the best analytics tools can often fail to identify business objective.
CPG enterprises often assimilate data from different external sources (subsidiaries, channel partners or data aggregators), which inherently brings the issue around data contextualization.
For instance:

  • Altered product identification from original manufacturer designated IDs
  • Products attributed to different classifications across different value chain partners
  • Product information in a different language

Without data contextualization CPG companies won’t be able to derive meaningful insights in real time.
According to Nielsen, “Harmonization is an ongoing process that requires strict governance and constant monitoring.” But when the process isn’t in its place and is unable to draw from both local knowledge and global expertise, it delivers inaccurate information.

Deriving actionable insights from the harmonized data

The spiral of innovation in the digital era hasn’t been kind to the CPG companies. To top the already existing sales, promotional, supply-chain, and finance data; social media is a new entrant to add to the woes. And what becomes critical for CPG organizations now is how to manage, streamline sort, filter and make sense of these data mountains; making them easily graspable and insightful in real time.
As most CPG companies expand their foothold in developing markets they unlock new opportunities, and the challenge to capture critical downstream data in the distribution channel.