Debdoot Mukherjee is the Chief Data Scientist and Head of AI at Meesho, the Indian origin social commerce platform at the forefront of the boundaryless workplace model that became a norm in the aftermath of the Covid-19 pandemic. Upon completing his postgraduate degree from IIT-Delhi, Mukherjee began his career in the research division at IBM, where he attained expertise in Information Retrieval and Machine Learning techniques. He then journeyed on to work in impactful roles at companies like Hike, Myntra and ShareChat before leading the AI and data science division at Meesho. In an exclusive interview with Analytics India Magazine, Debdoot Mukherjee opened up about his journey into data science, machine learning and everything AI. AIM: What attracted you to this field?

A brand-new area of computer science was initially referred to as "artificial intelligence" in 1955. Many daily jobs are being replaced by artificial intelligence, requiring less human involvement. But what precisely is this new AI technology? AI refers to the process of teaching a computer system to function and think like a human brain. This is often accomplished through reinforcement learning, in which the computer learns from past errors and observed patterns.

Looking at the latest TRDR Pocket video games console, you would never think it hosted one of the most sophisticated AI voice assistants on the market today. The £399 Gameboy-like device, which is the second handheld console developed by British company Go Games, has a boxy, retro appearance, with a shiny aluminium body and a 3.5-inch touchscreen display. It runs Android, giving users access to over 600,000 games and apps from the Google Play Store, including Fortnite and Call of Duty. The original TRDR Pocket, which was released in 2021 in partnership with American rapper and social media influencer Soulja Boy, shipped over 100,000 units. However, the reviews were far from glowing - with some pointing out it was almost identical to the Retroid Pocket, while others complained that the small touch screen made it awkward to type text and play certain games like shooters.

For decades, science fiction authors, futurists, and movie makers alike have been predicting the amazing (and sometimes catastrophic) changes that will arise with the advent of widespread artificial intelligence. So far, AI hasn't made any such crazy waves, and in many ways has quietly become ubiquitous in numerous aspects of our daily lives. From the intelligent sensors that help us take perfect pictures, to the automatic parking features in cars, to the sometimes frustrating personal assistants in smartphones, artificial intelligence of one kind of another is all around us, all the time. While we've yet to create self-aware robots like those that pepper popular movies like 2001: A Space Odyssey and Star Wars, we have made smart and often significant use of AI technology in a wide range of applications that, while not as mind-blowing as androids, still change our day-to-day lives. One place where artificial intelligence is poised to make big changes (and in some cases already is) is in education.

Multi-types of behaviors (e.g., clicking, adding to cart, purchasing, etc.) widely exist in most real-world recommendation scenarios, which are beneficial to learn users' multi-faceted preferences. As dependencies are explicitly exhibited by the multiple types of behaviors, effectively modeling complex behavior dependencies is crucial for multi-behavior prediction. The state-of-the-art multi-behavior models learn behavior dependencies indistinguishably with all historical interactions as input. However, different behaviors may reflect different aspects of user preference, which means that some irrelevant interactions may play as noises to the target behavior to be predicted. To address the aforementioned limitations, we introduce multi-interest learning to the multi-behavior recommendation. More specifically, we propose a novel Coarse-to-fine Knowledge-enhanced Multi-interest Learning (CKML) framework to learn shared and behavior-specific interests for different behaviors. CKML introduces two advanced modules, namely Coarse-grained Interest Extracting (CIE) and Fine-grained Behavioral Correlation (FBC), which work jointly to capture fine-grained behavioral dependencies. CIE uses knowledge-aware information to extract initial representations of each interest. FBC incorporates a dynamic routing scheme to further assign each behavior among interests. Additionally, we use the self-attention mechanism to correlate different behavioral information at the interest level. Empirical results on three real-world datasets verify the effectiveness and efficiency of our model in exploiting multi-behavior data. Further experiments demonstrate the effectiveness of each module and the robustness and superiority of the shared and specific modelling paradigm for multi-behavior data.

Online bipartite matching is a fundamental problem in online algorithms. The goal is to match two sets of vertices to maximize the sum of the edge weights, where for one set of vertices, each vertex and its corresponding edge weights appear in a sequence. Currently, in the practical recommendation system or search engine, the weights are decided by the inner product between the deep representation of a user and the deep representation of an item. The standard online matching needs to pay $nd$ time to linear scan all the $n$ items, computing weight (assuming each representation vector has length $d$), and then decide the matching based on the weights. However, in reality, the $n$ could be very large, e.g. in online e-commerce platforms. Thus, improving the time of computing weights is a problem of practical significance. In this work, we provide the theoretical foundation for computing the weights approximately. We show that, with our proposed randomized data structures, the weights can be computed in sublinear time while still preserving the competitive ratio of the matching algorithm.

Advanced technology has made it possible to run the world from the fingertips. Still, we humans thrive on more comfort and convenience in everything. Secure transactions that are hassle-free and quick are one of our most primary concerns. And with Paytalk Application, it is possible to fulfil all these demands. Various applications in the market promise and deliver features to aid the online transaction, and those applications require saving and careful data entry every single time.

Empower your kids to learn the basic concepts in Artificial Intelligence curated by experts from University of Oxford, IIT, MIT and Graz University of technology. Deep dive into one of the best AI Coding courses using Scratch from MIT, Snap from UC Berkeley, Phiro Code, Python and JavaScript. In a rapidly changing world, we make sure your child learns experientially with PBL activities & projects from the best in the industry. AI Programming courses for kids, Python coding classes for kids, Scratch coding for kids, Summer camp for kids

According to a recent McKinsey survey of executives, companies have pushed the time frame for digitizing many aspects of their business, from internal operations to supply chain and customer interactions, by three to four years. Digital products in those companies' portfolios have also shot some seven years ahead of where they had expected to be prior to the pandemic. The Great Resignation, skill shortages, supply chain disruptions, working from home, touchless customer experience, and agile process redesigns are paradigm shifts that businesses have rapidly needed to adapt to. But how do companies stay dynamic, resilient, and efficient in this new era? We believe that a big part of the solution may be found in digital employees, powered by automation and AI.

There has been an increased interest in applying machine learning techniques on relational structured-data based on an observed graph. Often, this graph is not fully representative of the true relationship amongst nodes. In these settings, building a generative model conditioned on the observed graph allows to take the graph uncertainty into account. Various existing techniques either rely on restrictive assumptions, fail to preserve topological properties within the samples or are prohibitively expensive for larger graphs. In this work, we introduce the node copying model for constructing a distribution over graphs. Sampling of a random graph is carried out by replacing each node's neighbors by those of a randomly sampled similar node. The sampled graphs preserve key characteristics of the graph structure without explicitly targeting them. Additionally, sampling from this model is extremely simple and scales linearly with the nodes. We show the usefulness of the copying model in three tasks. First, in node classification, a Bayesian formulation based on node copying achieves higher accuracy in sparse data settings. Second, we employ our proposed model to mitigate the effect of adversarial attacks on the graph topology. Last, incorporation of the model in a recommendation system setting improves recall over state-of-the-art methods.