A naukri.com initiative
Data Science News
Medium
1M
58
Image Credit: Medium
I 101: Breaking Down Artificial Intelligence
- Artificial intelligence (AI) is a field that enables computers to learn from data and make decisions without explicit programming.
- Machine learning is a subset of AI that uses statistical approaches to find patterns in data and improve performance over time.
- Deep learning is a subset of machine learning that simulates the functioning of the human brain to process complex patterns of data using artificial neural networks with multiple layers.
- The project lifecycle for an AI application involves problem definition, data acquisition and cleaning, model selection and training, model evaluation and refinement, deployment, and ongoing ML Ops to keep the model updated and improve performance.
Read Full Article
3 Likes
Analyticsindiamag
1M
366
Image Credit: Analyticsindiamag
Perplexity Announces Voice Assistant for iPhone
- Perplexity announced a new voice assistant for iOS, available through an update for the Perplexity app on the App Store.
- The assistant uses web browsing and multi-app actions to perform tasks like booking reservations, sending emails, playing media, and more.
- Users can customize the Action Button on their iPhone to use the assistant without opening the app.
- Perplexity is in talks with Motorola and Samsung to integrate its assistant on their devices, potentially becoming the default option or pre-installed app.
Read Full Article
22 Likes
Analyticsindiamag
1M
285
Image Credit: Analyticsindiamag
Persistent Systems Posts 25% Q4 Revenue Growth, Outpaces Big Indian IT Firms
- Persistent Systems reported a 25% YoY rise in consolidated net profit and revenues for Q4 FY25, outperforming larger Indian IT firms.
- The company recorded a net profit of ₹395.76 crore and revenue of ₹3,242 crore in the March quarter.
- Persistent Systems reported a 4.5% sequential growth in constant currency revenue, surpassing analyst projections.
- The company aims to achieve $2 billion in annual revenue by FY27.
Read Full Article
16 Likes
Analyticsindiamag
1M
203
Image Credit: Analyticsindiamag
Databricks to Invest $250 Million in India, Plans 50% Workforce Expansion
- Databricks announces a $250 million investment in India over the next three years, focusing on data and AI innovation.
- The company plans to expand its workforce in India by over 50%, reaching more than 750 employees by the end of the year.
- Databricks introduces the India Data + AI Academy to train 500,000 professionals and offers certifications upon completion.
- The company will open a new R&D office in Bengaluru and hire over 100 engineers to strengthen its research efforts.
Read Full Article
4 Likes
Analyticsindiamag
1M
143
Image Credit: Analyticsindiamag
Microsoft’s 365 Copilot Receives New Features in a Major Update
- Microsoft announced an update to the Copilot 365 application, introducing new features like AI-powered search and Create feature for image generation.
- The app now defaults to a chat-based interface and includes a new Agent Store.
- Microsoft Researcher and Analyst agents are now available for integration, providing deep research and data science capabilities.
- Microsoft also introduced Copilot Notebooks for real-time insights and audio summaries, along with memory and personalization features.
Read Full Article
8 Likes
Analyticsindiamag
1M
187
Image Credit: Analyticsindiamag
OpenAI Brings Image Generation Model to API via gpt-image-1
- OpenAI has released the gpt-image-1 multimodal image generation model to its API, allowing integration of image creation capabilities into tools and platforms.
- The model supports image generation from text prompts, detailed editing, and accurate rendering of styles and text.
- Companies such as Adobe, Airtable, Figma, Quora, and Wix are already using the gpt-image-1 model.
- The gpt-image-1 API includes safety features, pricing is based on token usage, and the model is now available globally via the Images API.
Read Full Article
10 Likes
Analyticsindiamag
1M
96
Image Credit: Analyticsindiamag
MeitY to Back Sarvam AI, Soket AI Labs, Gnani.ai for IndiaAI Mission
- The electronics and information technology ministry (MeitY) has selected Sarvam AI, Soket AI Labs, Gnani.ai, and Gan.ai as the first four shortlisted startups for building Indian AI models as part of the ₹10,000 crore IndiaAI Mission project.
- Soket AI Labs aims to develop a 120-billion parameter open-source Indic LLM while Gnani.ai and Gan.ai have proposed to build smaller language models.
- Sarvam AI is working on a 70-billion parameter multimodal AI model that supports Indian languages and English, and has already begun the development.
- MeitY is expected to formally announce the selected firms soon, with Union Minister Ashwini Vaishnaw likely to make the official declaration.
Read Full Article
5 Likes
Medium
1M
199
Image Credit: Medium
From Dashboards to Decisions: Building AI-First Data Products as a PM
- Traditional dashboards often fall short in delivering true business value, requiring users to interpret data.
- AI-first data products actively interpret data, surface insights, and recommend actions.
- Building AI-first data products requires PMs to expand their skill sets and responsibilities.
- Success in AI-first products requires balancing technical capabilities with human needs.
Read Full Article
11 Likes
Analyticsindiamag
1M
439
Image Credit: Analyticsindiamag
These Drones are Flying at 18,000 Ft, Above the Himalayas for the Indian Army
- Drone technology in India has rapidly evolved from novelty to necessity, being used in various critical operations.
- Regulatory bodies like DGCA are streamlining processes to support the growing drone ecosystem in India.
- Scandron, a Bengaluru-based manufacturer, has received DGCA certification for logistics drones and is leading in high-altitude logistics.
- Scandron drones have successfully delivered supplies to Indian Army troops in remote Himalayan terrains.
- These drones have replaced traditional methods in various sectors, from military logistics to agriculture and infrastructure inspection.
- Scandron drones have flown at 18,000 feet above the Himalayas with heavy payloads, providing critical supplies to inaccessible areas.
- The company has a dedicated high-altitude testing facility and focuses on designing drones for severe weather conditions and high altitudes.
- Beyond defence, Scandron drones are part of schemes empowering women farmers in efficient crop spraying.
- Scandron also supports industries like solar parks by using drones for transportation of solar panels.
- The company emphasizes practical AI applications in drones for safety and data analysis rather than autonomous warfare.
Read Full Article
26 Likes
Medium
1M
393
Image Credit: Medium
From BoW to GPT: A Brief History of Language Models and Where They’re Going
- Language models have evolved from simple word counters to billion-parameter models, built on decades of breakthroughs.
- The progress from Bag-of-Words (BoW) to n-gram models improved context prediction but lacked understanding.
- Word2Vec introduced word embeddings, representing words as vectors in a relational space.
- Recurrent Neural Networks (RNNs) and Long Short-Term Memory (LSTM) networks improved language sequence processing.
- The introduction of Transformers in 2017 revolutionized language modeling with self-attention.
- Models like BERT, GPT-2, and T5 are built on the Transformer architecture, focusing on context and relationships.
- Recent advancements have led to massive models like GPT-3 with 175 billion parameters and instruction-tuned models for various tasks.
- Future directions include domain-specific models, neuro-symbolic systems, and connecting language models with real-world data.
- The evolution of language models reflects our evolving understanding of language, culture, and the nuances of communication.
- The future of language models may prioritize smarter, more understanding models over simply scaling up in size.
- Understanding the history of language models is essential in grasping how each step contributes to the evolving capabilities of machines in language processing.
Read Full Article
23 Likes
Towards Data Science
1M
188
Exporting MLflow Experiments from Restricted HPC Systems
- Running MLflow experiments on restricted HPC systems can be challenging, as outbound TCP connections are often limited.
- To bypass direct communication limitations, a workaround involves setting up a local MLflow server with local directory storage.
- Steps include creating a virtual environment, installing MLflow, and using mlflow-export-import for data transfer.
- Exporting experiment data to a local temp folder on the HPC and transferring it to the remote MLflow server is crucial.
- Installation of Charmed MLflow (MLflow server, MySQL, MinIO) using juju on MicroK8s localhost is recommended.
- Prerequisites include Python 3.2 loaded on both HPC and MLflow server, with specific configuration settings.
- Issues like thread utilization errors can occur, but setting thread limits and environment variables can help in resolving them.
- The process involves exporting experiments, transferring runs to the MLflow server, importing data to MySQL and MinIO, and cleaning up ports.
- By spinning up a local MLflow server, exporting, and importing experiments, users can track and manage experiments in restricted environments.
- Security precautions, such as secure transfer methods and monitoring local storage, are crucial when implementing this workaround.
Read Full Article
11 Likes
Towards Data Science
1M
353
How to Benchmark DeepSeek-R1 Distilled Models on GPQA Using Ollama and OpenAI’s simple-evals
- The DeepSeek-R1 model gained attention for its reasoning abilities and cost-efficiency compared to other models.
- Assessing DeepSeek-R1's reasoning abilities programmatically offers deeper insights.
- Distilled models from DeepSeek-R1, varying in size, aim to replicate the larger model's performance.
- Distillation transfers reasoning abilities to smaller, more efficient models for complex tasks.
- The selection of a distilled model size depends on hardware capabilities and performance needs.
- Benchmarks like GPQA-Diamond are used to evaluate reasoning capabilities in LLMs.
- Tools like Ollama and OpenAI's simple-evals assist in evaluating reasoning models.
- Evaluation results of DeepSeek-R1's distilled model on GPQA-Diamond highlighted some challenges.
- Setting up Ollama and simple-evals for benchmarking involves specific configurations.
- Although distilled models may have limitations in complex tasks, they offer opportunities for efficient deployment.
Read Full Article
20 Likes
Medium
1M
91
From Zero to Data Science Portfolio in 4 Hours? How AI Transformed My Workflow
- Mito AI, an extension for Jupyter Lab, transformed the author's data science workflow.
- It accelerated the process of building a data analytics portfolio piece from scratch.
- The initial phase that typically takes a week was completed quickly with Mito AI.
- By automating coding tasks, it allowed the author to focus on the analysis itself.
Read Full Article
5 Likes
Medium
1M
298
Image Credit: Medium
Understanding N-Grams in NLP: Capturing Context Beyond Single Words
- N-grams capture word order and context in text, enhancing language understanding.
- They are valuable in improving NLP models' performance.
- Python and scikit-learn make it easy to extract N-grams.
- N-grams add important context to language structure in NLP models.
Read Full Article
17 Likes
Medium
1M
414
Image Credit: Medium
Paper Insights: Tree-of -Thoughts: Deliberate Problem Solving with Large Language Models
- The Tree-of-Thoughts (ToT) methodology allows language models to make decisions by exploring different reasoning paths and evaluating their progress.
- ToT is inspired by a problem-solving concept from the 1950s and uses a tree structure to represent different thoughts and potential solutions.
- ToT involves thought decomposition, thought generation, state evaluation, and search algorithms to explore the problem space.
- Experiments on challenging tasks showed that ToT outperformed standard prompting and chain-of-thought prompting approaches.
Read Full Article
24 Likes
For uninterrupted reading, download the app