A naukri.com initiative
Data Science News
Analyticsindiamag
1w
231
Image Credit: Analyticsindiamag
Perplexity Announces Residency Program
- Perplexity, the AI search engine startup, has announced a residency program for recent graduates and early-stage career professionals.
- The program offers a three-month full-time opportunity in San Francisco to gain experience in the AI ecosystem.
- Participants will work with the product team and develop new features for Perplexity.
- Outstanding performers may receive a full-time position at the end of the program.
Read Full Article
13 Likes
Analyticsindiamag
1w
176
Image Credit: Analyticsindiamag
How Warner Bros Discovery is Shaking Up Ads Using AI
- Warner Bros Discovery (WBD) is deepening roots in AI innovation, with the opening of a new Global Delivery Centre in Hyderabad.
- WBD currently houses 500 employees in Hyderabad and hopes to expand rapidly.
- WBD aims to personalise customer experiences through AI-driven data insights to enhance customer personalisation across its services.
- WBD CEO David Zaslav said, 'AI is going to have an increasing impact on society and our industry and we intend to take full advantage.'
- Shop with Max and Moments are two AI-based solutions aimed at making ads more relevant, engaging and useful on WBD's streaming platform, Max.
- Advanced contextual targeting, tailoring ads to match the emotional tone of content and seamless shopping during shows are two new ad formats added by WB Discovery.
- Beyond shoppable and contextual ads, WBD is introducing advanced targeting tools and interactive video formats.
- According to a Dentsu Asia Pacific global ad spend forecast, by 2027, nearly 80% of ad spend will be algorithm-driven.
- AI tools are being used by Kimberly Clark, Coca-Cola, Cadbury, Heinz, and Unilever to generate customised ads and match ads with audience moods, interests, and behaviours.
- WBD has named Anish Agarwal as its new VP of AI & automation to spearhead the development and execution of advanced AI and automation strategies.
Read Full Article
10 Likes
Analyticsindiamag
1w
403
Image Credit: Analyticsindiamag
Google Names Preeti Lobana Country Manager, VP for India
- Preeti Lobana has been appointed as the new country manager and vice president for Google India.
- In her new role, Lobana will focus on driving business growth and fostering digital inclusion using Google's technological innovations.
- Lobana highlights the potential of AI to unlock opportunities, increase productivity, and address critical challenges.
- Roma Datta Chobey will continue her leadership as managing director for Google India's Digital Native Industries, Media and Entertainment, and Government.
Read Full Article
24 Likes
Analyticsindiamag
1w
113
Image Credit: Analyticsindiamag
Composable Architectures Are Non-Negotiable
- Composable architectures provide a framework for building scalable AI-powered applications, which enables developers to create customized solutions more quickly.
- This modular approach provides flexibility, scalability, and enhanced control over the deployment of AI solutions.
- Scalability is one of the main benefits of composable architectures, as individual modules can be scaled independently based on demand.
- Composable architecture enables rapid experimentation and granular control, which are especially useful in the fast-paced world of generative AI.
- Foundation models serve as fundamental building blocks in composable generative AI systems, and they can be fine-tuned or augmented for specific tasks.
- The composable architecture excels in adaptability, and organizations can improve or replace individual components as technology evolves.
- Managing and optimizing prompts at scale poses significant challenges, which composable architecture addresses by treating prompt management as a separate module within the overall system.
- Composable architectures provide increased flexibility, but they also present unique security and compliance challenges.
- Implementing composable architectures can be challenging, and integrating multiple modules presents challenges, especially as AI tools and technologies advance rapidly.
- By embracing composable architectures, organizations can position themselves to adapt swiftly to AI’s evolving landscape, benefiting from the enhanced flexibility, scalability, and security that modular systems provide.
Read Full Article
6 Likes
Analyticsindiamag
1w
210
Image Credit: Analyticsindiamag
Microsoft Solves the Problem of LLM Data Scarcity
- Microsoft has shown off its breakthrough Phi-4 AI language model, which includes 14bn parameters and use of synthetic data for high-performance reasoning and problem solving. It is able to provide top level performance without the need for big data input and scaling, and also comes with native support for up to 10 Indian languages. Microsoft's approach to high-quality datasets over quantity is part of an emerging trend that sees evidence of untapped data either held in corporate vaults or not being used in easily digestible digital formats.
- Phi-4's synthetic data serves as a more effective mechanism for learning by using structured, diverse and nuanced datasets.
- Frontier models will have to use different techniques and methods, beyond increasing parameter counts, as data for training like the data already available might have reached its peak.
- Microsoft Phi-4 AI language model has surpassed many of its competitors.
- It is expected to have far-reaching impacts on countries where most people cannot afford top-of-the-line models.
- Microsoft's focus on synthetic datasets is a step away from the original scaling hypothesis, instead choosing to prioritise the quality of datasets over their quantity.
- Frontier models have been a hot topic recently, with AI researchers contemplating the end of pre-training.
- Untapped and locked data sources may also be a means to improve machine learning models.
- Microsoft's new language model packs plenty of promise.
- Phi-4's use of synthetic data has brought about a paradigm shift to the earlier assumption is that the more data used, the better.
Read Full Article
12 Likes
Medium
1w
235
Image Credit: Medium
Working with Graphs in SciPy
- Graphs are essential structures in various computational fields, including computer science, data analysis, and network theory.
- SciPy provides tools to work with graphs through its scipy.sparse module.
- Sparse matrices are often used to represent graphs, especially when most possible edges between nodes are absent.
- A graph can be represented by an adjacency matrix, where each non-zero value represents an edge between two nodes.
Read Full Article
14 Likes
Medium
1w
378
Image Credit: Medium
The Digital World and the Evolution Toward Web 3.0
- Web 3.0 integrates blockchain technology to decentralize data storage and usage, enhancing data privacy and reducing data breach risks by up to 70% compared to traditional systems.
- Self-sovereign identity (SSI) uses blockchain to provide a secure, tamper-proof digital identity and can reduce identity theft incidents by nearly 90%, signaling a paradigm shift in how digital privacy is managed.
- Web 3.0 introduces new economic models through blockchain-based platforms that enable direct interactions between users, removing intermediaries, and creating fairer revenue distribution systems.
- Decentralized autonomous organizations (DAOs) empower members to participate in decision-making and share in the platform’s revenue fostering a participatory and equitable digital economy.
- Blockchain’s immutability ensures that once data is recorded, it cannot be altered or deleted, reducing the risk of fraud and manipulation and makes them 98% more resilient to cyberattacks than traditional systems.
- As Web 3.0 evolves, it creates new roles and opportunities in technology, art, and education, such as blockchain development, decentralized finance, NFT marketplaces, education, and consulting.
- Web 3.0 is not without challenges, high energy consumption, regulatory uncertainty, and accessibility barriers being a few significant hurdles towards building a decentralized future.
- The decentralization of data raises ethical questions, such as the proliferation of illegal content on decentralized platforms, which can be mitigated by striking a balance between freedom and accountability.
- Realizing Web 3.0's vision requires collaboration among technologists, policymakers, and users to overcome technical and ethical challenges, and the potential benefits are a freer, fairer, and more secure digital world.
- By embracing this transformation, we can reclaim control over our digital lives and unlock unprecedented opportunities for innovation and collaboration.
Read Full Article
22 Likes
Medium
1w
21
How AI Became the Ultimate Life Hack Let's face it: life has gotten a whole lot easier thanks to…
- AI is revolutionizing shopping by providing personalized recommendations and chatbot assistance.
- Health tech powered by AI monitors fitness levels, sleep patterns, and offers mental health support.
- Smart homes equipped with AI automate tasks like adjusting lights, temperature, and brewing coffee.
- AI-driven apps like Google Maps and Waze simplify travel by providing real-time traffic updates and flight price analysis.
Read Full Article
1 Like
Medium
1w
63
Understanding Tensors: The Backbone of Modern AI
- Tensors are a data structure used in AI and data science.
- They are a unified way to store and process data of any dimensionality.
- Tensors combine scalars, vectors, and matrices into a single structure.
- Popular AI libraries like TensorFlow and Scikit-learn rely on tensors.
Read Full Article
3 Likes
Analyticsindiamag
1w
63
Image Credit: Analyticsindiamag
BITS Pilani’s ‘Zero Attendance Policy’ Has a Lesson for All Indian Universities
- BITS Pilani has been successfully running its Zero Attendance Policy since 1946.
- Under this policy, students are not required to attend classes but are evaluated entirely on the basis of exams and assessments.
- This policy is used to motivate students to pursue entrepreneurial paths and build start-ups for the country.
- Leading universities like IITs and DIT University have also incorporated the initiative of temporarily withdrawing programs.
- Unicorns like Swiggy, redBus, and Groww were born out of BITS Pilani.
- Though this policy exists since five decades at BITS, this has sparked a debate on social media platforms whether such an approach would work for other institutes.
- For the Zero Attendance Policy to be successful, the institution has to ensure that students are always prepared for the courses throughout the semester and that attendance isn't a criterion for appearing for exams.
- The policy also ensures that faculty have to be innovative and attract students to attend their classes and lectures regularly.
- Manoj BS, Professor at IIST, said that the policy can be effective only if appropriate assessment is done and mandating a certain percentage of attendance is essential.
- The freedom of not attending classes may not work well in most institutions, but for BITSian, it has demonstrably worked where students are exposed to many opportunities through this policy.
Read Full Article
3 Likes
VentureBeat
1w
71
Image Credit: VentureBeat
We’ve come a long way from RPA: How AI agents are revolutionizing automation
- AI agents are revolutionizing automation, replacing traditional automation tools like RPA.
- AI agents, capable of autonomous thinking and collaboration, are the next wave of intelligent automation.
- Vertical AI agents are specialized for specific industries, bringing new possibilities to enterprise automation.
- Achieving high accuracy and investing in evaluation frameworks are crucial for successful AI agent deployment.
Read Full Article
4 Likes
Medium
1w
105
Image Credit: Medium
Gen AI for Business Newsletter Edition # 35
- Google and Microsoft are among the companies dominating headlines this week with leading edge tool, focused on AI and ML technology, email summaries, voicemail transcription, virtual try-ons, AI-generated quizzes, and creative tools such as Magic Eraser and Pixel Studio, enhancing learning, planning, and visual experiences.
- According to surveys, AI features are a low priority for smartphone owners, with 61% prioritizing privacy concerns deterring adoption despite Gen Z and Millennial enthusiasm.
- Generative AI is outpacing the internet and PCs, with 28% of workers using AI at their jobs, transforming sectors like telecom and finance with promising operational improvements.
- Kerry Sheehan, a leading figure in AI regulation for the UK, advocates for continuous learning and collaboration, and encouraging women in AI to support inclusion and representation through networking, coaching, and active participation in AI development.
- NVIDIA is opening its first AI-focused R&D center in Vietnam, leveraging local STEM talent to drive innovation across industries while collaborating with over 100 startups and universities.
- Capital One’s survey reveals a disconnect between executives’ belief in AI readiness and IT struggles with data quality, emphasizing the need for clean data and use-case-driven approaches to align AI goals effectively.
- NYC leverages AI for cybersecurity, filtering billions of events into actionable insights, while enhancing public services with transparent municipal AI strategies; in HR, companies like Walmart and Siemens use generative AI for personalized training, automated recruitment, and predictive insights to improve employee and customer experiences.
- Gartner predicts AI will handle 15% of daily decisions by 2028, while Deloitte reports AI is revolutionizing IT by automating coding, modernizing infrastructure, and reshaping IT into an enabling function, despite talent shortages and cybersecurity risks.
- Google Cloud’s generative AI tools enable businesses to enhance productivity and creativity with no-code options like Vertex AI, while researchers are leveraging LLMs to reduce stigma in substance use discourse through empathetic language suggestions.
- OpenAI’s new features, Canvas and Projects, streamline collaboration and organization, offering seamless spaces for refining tasks and managing projects efficiently.
Read Full Article
6 Likes
Dev
1w
396
Image Credit: Dev
Big O Notations
- Big O Notation is a method to measure how fast an algorithm is running.
- Temporal Complexity determines how long an algorithm takes to execute relative to the input size.
- Spatial Complexity determines how much memory is allocated to find the item we need.
- Big O Notation helps in determining how scalable an algorithm is.
- The execution time of an algorithm is denoted using Big O Notation.
- Temporal constant with O(1) defines the operations that take a constant execution time.
- Linear time with O(n) defines that execution time increases in proportion to the size of an array.
- Logarithmic time with O(log n) means input size increases linearly, however, execution time increases logarithmically.
- Linearithmic/quasilinear time with O(n log n) is a moderately growing time complexity that is implied while performing logarithmic operations n times.
- Quadratic time O (n²) is when the execution time increases quadratically with the number of inputs. It generally happens when reading a matrix or when nested loops are present.
Read Full Article
23 Likes
Medium
1w
257
Image Credit: Medium
Understanding the Differences Between NaN vs Null vs None Values in Python
- The difference between NaN, None, and null in python and when to use each
- In pure Python, you can assign a string a value or set it as empty '' or None.
- Null is not an option… you can choose to set it as 'null' but there could be some issues down the road with this (because it doesn’t evaluate to null in Python).
- Numpy’s np.nan value actually evaluates to the ‘float’ class here, and surprise surprise, it evaluates to Boolean value of True.
- Test cases have been done for each of the Empty string, None, NaN, Zero values to see how python treats each of these values.
- While summing a list of values (which include an np.nan value) doesn’t throw an exception… it returns nan! It does not exclude the nan value (which is probably what you want to do).
- Adding (or multiplying) an np.nan value by anything results in, you guessed it, an nan value again.
- np.nansum(my_list) will ignore np.nan values and treat them as 0’s in summation so there are no errors generated.
- At this point, my brain is fried and I want it to absorb what I’ve learned.
- For a production codebase, it is better to simply use python's None.
Read Full Article
15 Likes
Medium
1w
151
Image Credit: Medium
The History of AI
- AI originated from ancient myths and stories, but took shape in the 20th century.
- Alan Turing posed the question of whether machines can think, leading to the creation of the Turing Test.
- AI saw progress in the 1960s and 1970s, but faced setbacks in the AI Winter of the 1970s and 1980s.
- Expert systems and the rise of machine learning revived AI in the 1980s and 2000s respectively.
- AI gained prominence in the 2010s with advancements in big data and cloud computing.
- The future of AI involves its pervasive presence and challenges in ethics and regulation.
Read Full Article
9 Likes
For uninterrupted reading, download the app