Vinayak Mane

Happy to share that our paper, "UniMTS: Unified Pre-training for Motion Time Series", has been accepted to the Neural Information Processing Systems (NeurIPS) 2024. This is the first pre-trained model specifically designed for human motion time series, excelling in zero-shot activity recognition across a wide range of wearable devices, regardless of body placement or orientation. Our approach follows a CLIP-inspired pre-training framework, where we leverage contrastive learning to align motion data with LLM-enhanced activity descriptions. https://lnkd.in/gjJcVSiR #NeurIPS2024 #neurips #neurips2024 #neurips24 #MachineLearning

92

9 Comments

How are Graph Convolutional Neural Networks (GCNNs) a generalisation of Convolutional Neural Networks (CNNs)? My next blog in the Graph series on "Deep Learning with Graphs" is on the way (well, it's already written on paper) and it's going to be a blockbuster. The beauty of Graph Machine Learning is that the more you read about it, the more you feel that your neurons (real ones!) are learning to think better! And you're already in love with it! Coming to the point: One might recall the convolution operations in a CNN. You have a grid of pixels representing an image. You have a 3 x 3 filter. You slide the filter over the image and carry out aggregation to obtain a new pixel value. You proceed in the same fashion over the whole grid, sliding towards the right , bottom, carrying out the convolution operation and finally getting a new transformed, reduced image through the series of aggregations. You repeat this through the different convolutional layers which together form a Convolutional Neural Network Now think of a Graph. It is not simple grid like an image. The topology of a Graph is arbitrary. There is no notion of left, bottom, right in a Graph like in an image which is represented through a grid of pixels. You cannot slide over a Graph as simplistically you did and performed the convolutional operations in an image. How do you approach the problem? How do you carry out convolutional operations in a Graph? Now take a node in the Graph. Take all the nodes which are the immediate neighbours of this node. The convolutional operation here is: You aggregate the information from all the neighbours and pass it to the node. The information is nothing but the emedfings of the neighbour nodes. You add together the emedding of the node under consideration. You transform the aggregated embedding by multiplying with a weight Matrix and transform the node's own embedding by multiplying with a different weight matrix. This is the typical linear transformation operation in a NN. You apply non linearity - the activation function like ReLu. Now think about a CNN taking the same grid and considering a central pixel. Consider the rest of the pixels around the central pixel. You are essentially doing the same thing: taking information from the neighbour pixels along with the central pixel, aggregating, as I described above for the node in the Graph. Hence a Graph Convolutional Network is a generalisation of CNN The story goes, you do the same for every node on the Graph. Each node in the Graph represented through a different neural network architecture! This is amazing 🔥🔥🔥💓💓💓 #graphs #machinelearning #deeplearning My Technical Blogs Library: 1) https://lnkd.in/ejyWacXn 2) https://lnkd.in/bfrkRjq Follow Ajay Taneja for insights on: Data Science, Data Engineering, Finite Element Analysis, Non Linear Finite Element Analysis, Fatigue, Fracture Mechanics

22

When it comes to vector operations, two concepts that are often confused with each other are Dot Product and Cosine Similarity. While they may seem similar at first glance, the key difference lies in how they handle magnitude. 𝘋𝘰𝘵 𝘗𝘳𝘰𝘥𝘶𝘤𝘵: 𝘪𝘵 𝘮𝘦𝘢𝘴𝘶𝘳𝘦𝘴 𝘵𝘩𝘦 𝘦𝘹𝘵𝘦𝘯𝘵 𝘵𝘰 𝘸𝘩𝘪𝘤𝘩 𝘵𝘸𝘰 𝘷𝘦𝘤𝘵𝘰𝘳𝘴 𝘢𝘳𝘦 𝘱𝘰𝘪𝘯𝘵𝘪𝘯𝘨 𝘪𝘯 𝘵𝘩𝘦 𝘴𝘢𝘮𝘦 𝘥𝘪𝘳𝘦𝘤𝘵𝘪𝘰𝘯, 𝘤𝘰𝘯𝘴𝘪𝘥𝘦𝘳𝘪𝘯𝘨 𝘵𝘩𝘦𝘪𝘳 𝘭𝘦𝘯𝘨𝘵𝘩. 𝘊𝘰𝘴𝘪𝘯𝘦 𝘚𝘪𝘮𝘪𝘭𝘢𝘳𝘪𝘵𝘺: 𝘵𝘢𝘬𝘦𝘴 𝘵𝘩𝘦 𝘥𝘰𝘵 𝘱𝘳𝘰𝘥𝘶𝘤𝘵 𝘢𝘯𝘥 𝘥𝘪𝘷𝘪𝘥𝘦𝘴 𝘪𝘵 𝘣𝘺 𝘵𝘩𝘦 𝘱𝘳𝘰𝘥𝘶𝘤𝘵 𝘰𝘧 𝘵𝘩𝘦 𝘮𝘢𝘨𝘯𝘪𝘵𝘶𝘥𝘦𝘴 𝘰𝘧 𝘵𝘩𝘦 𝘵𝘸𝘰 𝘷𝘦𝘤𝘵𝘰𝘳𝘴. 𝗔 𝗣𝗿𝗮𝗰𝘁𝗶𝗰𝗮𝗹 𝗘𝘅𝗮𝗺𝗽𝗹𝗲 Let's consider three vectors representing sentences: s1 = [1, 2, 1] ("The cat sat on the mat.") s2 = [1, 1, 1] ("The cat is sitting on the mat.") s3 = [4, 5, 6] ("The cat is sitting comfortably on the warm mat in the living room near the fireplace.") Dot product between s1 and s2: 4 Dot product between s1 and s3: 20 (large difference due to magnitude) Cosine Similarity between s1 and s2: 0.942 Cosine Similarity between s1 and s3: 0.930 (smaller difference, focusing on direction) 𝗦𝗼, 𝘄𝗵𝗲𝗻 𝘀𝗵𝗼𝘂𝗹𝗱 𝘆𝗼𝘂 𝘂𝘀𝗲 𝗲𝗮𝗰𝗵? Dot Product is suitable for applications like search engines, where the relevance of a document to a query is determined by the frequency of key terms (longer vectors naturally have higher dot products). Other examples can be, in recommendation systems to measure similarity between users or between items etc. Cosine Similarity is ideal for comparing the similarity between documents, sentences, or word embeddings in NLP tasks, as it normalises for length and focuses on direction. Other examples can be, analyzing connections or interactions in social networks, retrieving similar images based on feature vectors etc. #DotProduct #CosineSimilarity #VectorOperations #SearchEngines #NLP #DocumentSimilarity #SentenceSimilarity #SocialNetworkAnalysis #ImageRetrieval #MachineLearning #DataScience #Mathematics #Algorithms #DataAnalysis

762

33 Comments

Recently, I have taken a session in the Mathematical for Machine Learning series which is an initiative taken by the TensorFlow User Group Ghaziabad (TFUG Ghaziabad). In this session we discussed "Dimensionality Reduction - PCA's Mathematical Opus", where we explored real world applications of finding planets and investigating crimes. The session was also focused on the math behind PCA and implemented it. 🔴Recording of the Session - https://lnkd.in/dHtSubWZ 🗃️Repo Resources - https://lnkd.in/dWC7_xRz

43

2 Comments

📚 Book Review: "Mastering PyTorch" – Create and deploy deep learning models from CNNs to multimodal models, LLMs, and beyond by Ashish Ranjan Jha 🚀 I recently finished reading "Mastering PyTorch," and I must say it is an excellent resource for anyone looking to delve deep into the world of deep learning using PyTorch. Whether you are a data scientist, machine learning researcher, or deep learning practitioner, this book offers a wealth of knowledge and practical insights to enhance your skills and understanding. 🔑 Key Highlights: 1️⃣ Comprehensive Overview: The book kicks off with an introduction to deep learning and PyTorch, setting a strong foundation for beginners and experienced practitioners alike. 2️⃣ Advanced CNN and RNN Architectures: It covers SOTA advancements in Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs), providing hands-on examples and practical applications. 3️⃣ Transformers and Hybrid Models: The exploration of transformers and hybrid models is particularly insightful, showcasing how these models have revolutionized natural language processing. 4️⃣ Graph Neural Networks (GNNs): A detailed dive into GNNs using PyTorch Geometric, explaining the concepts and implementations of Graph Convolutional Networks (GCNs) and Graph Attention Networks (GATs). 5️⃣ Generative Models: The sections on music and text generation, neural style transfer, GANs, and diffusion models are fascinating, demonstrating the creative potential of deep learning. 6️⃣ Deep Reinforcement Learning: Practical examples of training reinforcement learning agents in PyTorch add another dimension to the book's versatility. 7️⃣ Model Deployment: Step-by-step guidance on deploying PyTorch models into production, including mobile deployment for Android and iOS, is incredibly valuable for real-world applications. 8️⃣ Optimization and AutoML: Chapters on model training optimizations, distributed training, and AutoML provide techniques to efficiently train large models and automate the ML workflow. 9️⃣ Explainable AI: The focus on interpretability using tools like Captum ensures that the models are not just powerful but also understandable. 🔟 Integration with Hugging Face: Leveraging Hugging Face libraries with PyTorch to build cutting-edge AI models is a perfect ending to this comprehensive guide. If you have hands-on experience with Python and a basic understanding of PyTorch, this book is for you. It caters to those who wish to implement advanced deep learning paradigms and need practical examples to bridge the gap between theory and practice. 👨‍⚖️ "Mastering PyTorch" is more than just a technical guide; it is a gateway to mastering deep learning with one of the most powerful frameworks available today. 🌟 As an optim fan, Chapters 12, 13, 15 and 16 specially recommended! #DeepLearning #PyTorch #MachineLearning #AI #DataScience #BookReview #Transformers #ReinforcementLearning

26

5 Comments

NLP (LLM) Research Roadmap I’m happy to be an NLP research mentor for - PhD candidates without a good faculty supervisor - Industry professionals who want to gain research experience by publishing papers. Guidance will be provided via 1-to-1 paid Google Meet sessions. Interested people can DM me with all the relevant details. "Hi" or "incomplete messages" will be ignored due to time constraints. NLP research involves various phases like - Research Foundation (learning pre-requisites) - Research Papers Reading - Research Problem Identification - Research Gaps Identification - Research Idea Formulation - Research Idea Implementation - Research Paper Writing - Research Paper Submission and Review A good faculty supervisor provides proper guidance in every phase of NLP research. Unfortunately, some faculty supervisors will never do this which makes the journey of PhD candidates difficult. Due to a lack of proper guidance from the faculty supervisor, some of the PhD candidates despite being highly talented finish their PhD with average or below-average papers even after spending more than five years. #llms #nlproc #india #phdresearch #phd

255

5 Comments

🚨ACL 2024 Outstanding Paper Award Alert 🚨 ➡️Paper Title: INDICLLMSUITE: A BLUEPRINT FOR CREATING PRE-TRAINING AND FINE-TUNING DATASETS FOR INDIAN LANGUAGES 🌟Few pointers from the paper 🎯Despite the considerable advancements in English LLMs, the progress in building comparable models for other languages has been hindered due to the scarcity of tailored resources. 👁In this paper Authors work aims to bridge this divide by introducing an expansive suite of resources specifically designed for the development of Indic LLMs, covering 22 languages, containing a total of 251B tokens and 74.8M instruction-response pairs. 🎯Recognizing the importance of both data quality and quantity, their approach combines highly curated manually verified data, unverified yet valuable data, and synthetic data. 🎯They built a clean, open-source pipeline for curating pre-training data from diverse sources, including websites, PDFs, and videos, incorporating best practices for crawling, cleaning, flagging, and deduplication. 🎯For instruction-fine tuning, they amalgamate existing Indic datasets, translate/transliterate English datasets into Indian languages, and utilize LLaMa2 and Mixtral models to create conversations grounded in articles from Indian Wikipedia and Wikihow. 🎯Additionally, they addressed toxicity alignment by generating toxic prompts for multiple scenarios and then generated non-toxic responses by feeding these toxic prompts to an aligned LLaMa2 model. 🎯 They hope that the datasets, tools, and resources released as a part of this work will not only propel the research and development of Indic LLMs but also establish an open-source blueprint for extending such efforts to other languages. 🏢Organization: 🛎️Nilekani Centre at AI4Bhārat🛎️ Indian Institute of Technology, Madras 🛎️Microsoft🛎️Sant Longowal Institute of Engineering and Technology 🛎️Indian Institute of Information Technology Design & Manufacturing Kancheepuram, 🛎️Maharaja Agrasen Institute Of Technology, Delhi 🛎️Sarvam ,🛎️ NICT 🧙Paper Authors: Mohammed Safi Ur Rahman Khan, Priyam Mehta , Ananth Sankar , Umashankar Kumaravelan, Sumanth Doddapaneni, Suriyaprasaad G, Varun Balan, Sparsh Jain, Anoop Kunchukuttan, Pratyush Kumar, Raj Dabre, Mitesh Khapra 1️⃣Read the Full Paper here: https://lnkd.in/gqwQST-r 2️⃣Code: https://lnkd.in/giWPVntx 🎉 🎉  Congratulations to all the authors 🎉 🎉 Find this Valuable 💎 ? ♻️REPOST and teach your network something new Follow me 👣, Naveen Manwani, for the latest updates on Tech and AI-related news, insightful research papers, and exciting announcements. #ACL2024 #Indicllms #llms #llama2 #mixtral

15

For those who couldn't make it to SF in September for the AI Conference, check out the video link for the talk I gave on Graph RAG here 👇🏽. https://lnkd.in/gKR7AS4T The original BlackRock/Nvidia Hybrid RAG paper mentions triples (likely knowledge graphs using an RDF data model), and it's also light on implementation details of graph retrieval. Recently, I've been experimenting with strategies that utilize lightweight prompting frameworks and a reranker to combine results from graph + vector RAG, and this approach works surprisingly well even when using Kùzu's property graph data model and text2Cypher to retrieve from the graph. Give the video a watch first, and then check out the code here: https://lnkd.in/gaAWjqYv Happy to discuss more with anyone who has thoughts/comments on this methodology!

19

🌟 Just Completed the Ultimate PyTorch Course!  I'm thrilled to share that I've just completed an extensive PyTorch course from Zero To Mastery through Udemy! Here’s a snapshot of what I learned: 📚 1. PyTorch Fundamentals: Understanding tensors and how they are used in machine learning. 2. PyTorch Workflow: Steps from data to tensors to trained neural network models. 3. Neural Network Classification: Building models to classify data. 4. Computer Vision with PyTorch: Creating models for image recognition. 5. Custom Datasets: Loading and processing custom data. 6. Modular Code: Writing reusable Python scripts for machine learning. 7. Transfer Learning: Leveraging pre-trained models for new tasks. 8. Experiment Tracking: Keeping track of model performance. 9. Paper Replicating: Replicating cutting-edge research papers. 10. Model Deployment: Deploying models to the web. This particular course stood out to me for many reasons:   -Hands-On Projects: Running experiments, completing exercises, and building real-world models. -Comprehensive Coverage: From basics to advanced topics, everything is covered step-by-step. -Expert Instruction: Learning from a professional machine learning engineer with real-world experience. -Community Support: Access to an exclusive online community of students, alumni, mentors, TAs, and instructors. I'm excited to apply these skills in creating innovative AI solutions, and pursuing opportunities as a Machine Learning Engineer. The journey has just begun, and I'm ready to take on new challenges in this rapidly evolving field! #PyTorch #MachineLearning #DeepLearning #AI

25

2 Comments

Finding better entity embeddings, and capturing the most suited representation of an entity is an area of cutting-edge research. While working on improving the entity embeddings, here are the top areas/ideas one can think upon. Using a tremendous amount of data for training Using more sophisticated neural network architectures. Utilising the side information of the entity along with the context. A better Negative Sampling approach. Positive sample data is available but how does one create negative samples? And if negatives are too many how to better sample them? Utilising the available information on the sequence in which the entities interacted or were interacted with. And training sequence to sequence models around it. Here's a detailed video for Data Science enthusiasts I have created on the topic covering all the theory and a practical exercise training billions of rows to generate high-quality embeddings. Like and subscribe for more such interesting concepts. Also, like and share over here for maximum reach. : ) Video Link: youtu.be/DN4S96oHRhE YT channel Link:  youtube.com/@datatrek #datatrek #datascience #machinelearning #statistics #deeplearning #ai

18

Finding better entity embeddings, and capturing the most suited representation of an entity is an area of cutting-edge research. While working on improving the entity embeddings, here are the top areas/ideas one can think upon. Using a tremendous amount of data for training Using more sophisticated neural network architectures. Utilising the side information of the entity along with the context. A better Negative Sampling approach. Positive sample data is available but how does one create negative samples? And if negatives are too many how to better sample them? Utilising the available information on the sequence in which the entities interacted or were interacted with. And training sequence to sequence models around it. Here's a detailed video for Data Science enthusiasts I have created on the topic covering all the theory and a practical exercise training billions of rows to generate high-quality embeddings. Like and subscribe for more such interesting concepts. Also, like and share over here for maximum reach. : ) Video Link: youtu.be/DN4S96oHRhE YT channel Link:  youtube.com/@datatrek #datatrek #datascience #machinelearning #statistics #deeplearning #ai

9

🚨ECCV 2024 Paper Alert 🚨 ➡️Paper Title: T-MAE: Temporal Masked Autoencoders for Point Cloud Representation Learning 🌟Few pointers from the paper 🎯The scarcity of annotated data in LiDAR point cloud understanding hinders effective representation learning. Consequently, scholars have been actively investigating efficacious self-supervised pre-training paradigms. 🎯Nevertheless, temporal information, which is inherent in the LiDAR point cloud sequence, is consistently disregarded. 🎯To better utilize this property, authors of this paper proposed an effective pre-training strategy, namely “Temporal Masked Auto-Encoders (T-MAE)”, which takes as input temporally adjacent frames and learns temporal dependency. 🎯A SiamWCA backbone, containing a Siamese encoder and a windowed cross-attention (WCA) module, is established for the two-frame input. 🎯Considering that the movement of an ego-vehicle alters the view of the same instance, temporal modeling also serves as a robust and natural data augmentation, enhancing the comprehension of target objects. SiamWCA is a powerful architecture but heavily relies on annotated data. 🎯Their T-MAE pre-training strategy alleviates its demand for annotated data. Comprehensive experiments demonstrate that T-MAE achieves the best performance on both Waymo and ONCE datasets among competitive self-supervised approaches. 🏢Organization:  University of Amsterdam 🧙Paper Authors: Weijie Wei, Fatemeh Karimi Nejadasl, PhD, Theo Gevers, Martin R. Oswald 1️⃣Read the Full Paper here: https://lnkd.in/gKMuGJM5 2️⃣Project Page: https://lnkd.in/g5CC6a7H 3️⃣Code: https://lnkd.in/gHXXt6Gf 🎥 Be sure to watch the attached Presentation - Sound on 🔊🔊 Find this Valuable 💎 ? ♻️REPOST and teach your network something new Follow me 👣, Naveen Manwani, for the latest updates on Tech and AI-related news, insightful research papers, and exciting announcements. #ECCV2024 #lidarpointcloud #selfsupervisedlearning

24

I came across an excellent GitHub repository packed with hands-on Jupyter notebooks for anyone eager to explore LLMs. It covers: 🔹 Introduction to LLMs – Understanding the basics. 🔹 Tokens & Embeddings – The foundation of language modeling. 🔹 Model Internals – Peeking under the hood of LLMs. 🔹 Text Applications – Classification, clustering, and topic modeling. 🔹 Prompt Engineering – Crafting prompts for better performance. 🔹 Advanced Text Generation – Tools and techniques. 🔹 Semantic Search – Intelligent information retrieval. 🔹 Multimodal Models – Combining text and image understanding. 🔹 Customization – Building embeddings, fine-tuning BERT, and generative models. Big thanks to the repo creator for sharing such a valuable resource. Here is the repo: https://lnkd.in/gUKg7Jp7

77

2 Comments

🚨ECCV 2024 Paper Alert 🚨 ➡️Paper Title: UpFusion: Novel View Diffusion from Unposed Sparse View Observations 🌟Few pointers from the paper 🎯In this paper, authors have proposed UpFusion, a system that can perform novel view synthesis and infer 3D representations for an object given a sparse set of reference images without corresponding pose information. 🎯Current sparse-view 3D inference methods typically rely on camera poses to geometrically aggregate information from input views, but are not robust in-the-wild when such information is unavailable/inaccurate. 🎯 In contrast, UpFusion sidesteps this requirement by learning to implicitly leverage the available images as context in a conditional generative model for synthesizing novel views. 🎯They incorporated two complementary forms of conditioning into diffusion models for leveraging the input views:  a) via inferring query-view aligned features using a scene-level transformer,  b) via intermediate attentional layers that can directly observe the input image tokens. 🎯They showed that this mechanism allows generating high-fidelity novel views while improving the synthesis quality given additional (unposed) images. 🎯 They evaluated their approach on the Co3Dv2 and Google Scanned Objects datasets and demonstrate the benefits of their method over pose-reliant sparse-view methods as well as single-view methods that cannot leverage additional views. 🎯Finally, they also showed that their learned model can generalize beyond the training categories and even allow reconstruction from self-captured images of generic objects in-the-wild. 🏢Organization: Carnegie Mellon University, Snap Inc. Research 🧙Paper Authors: Bharath Raj N, Hsin-Ying Lee, Sergey Tulyakov, Shubham Tulsiani 1️⃣Read the Full Paper here: https://lnkd.in/gCpJhuFk 2️⃣Project Page: https://lnkd.in/gbv36iif 3️⃣Code: https://lnkd.in/gGW6wQ-Q Find this Valuable 💎 ? ♻️REPOST and teach your network something new Follow me 👣, Naveen Manwani, for the latest updates on Tech and AI-related news, insightful research papers, and exciting announcements. #ECCV2024

20

🚨Paper Alert 🚨 ➡️Paper Title: A Multimodal Automated Interpretability Agent 🌟Few pointers from the paper 🎯In this paper authors have described MAIA, “a  🤖 Multimodal Automated Interpretability Agent 🤖”. MAIA is a system that uses neural models to automate neural model understanding tasks like feature interpretation and failure mode discovery. 🎯It equips a pre-trained vision-language model with a set of tools that support iterative experimentation on subcomponents of other models to explain their behavior. 🎯These include tools commonly used by human interpretability researchers: for synthesizing and editing inputs, computing maximally activating exemplars from real-world datasets, and summarizing and describing experimental results. 🎯Interpretability experiments proposed by MAIA compose these tools to describe and explain system behavior. 🎯Authors evaluated  applications of MAIA to computer vision models. They first characterize MAIA’s ability to describe (neuron-level) features in learned representations of images. 🎯Across several trained models and a novel dataset of synthetic vision neurons with paired ground-truth descriptions, MAIA produces descriptions comparable to those generated by expert human experimenters. 🎯Then Authors showed that MAIA can aid in two additional interpretability tasks: reducing sensitivity to spurious features, and automatically identifying inputs likely to be mis-classified. 🏢Organization:  MIT Computer Science and Artificial Intelligence Laboratory (CSAIL) 🧙Paper Authors: Tamar Rott Shaham, Sarah Schwettmann, Franklin Wang, Achyuta Rajaram, Evan Hernandez, Jacob Andreas , Antonio Torralba 1️⃣Read the Full Paper here: https://lnkd.in/gjmN_gaR 2️⃣Project Page: https://lnkd.in/gxPr_jpM 3️⃣Experiment Browser: https://lnkd.in/gGe7FyUY 🎥 Be sure to watch the attached Demo Video-Sound on 🔊🔊 Music by Sergii Pavkin from Pixabay Find this Valuable 💎 ? ♻️REPOST and teach your network something new Follow me 👣, Naveen Manwani, for the latest updates on Tech and AI-related news, insightful research papers, and exciting announcements.

2

📚 Mind Versus Data in Machine Learning 📊 I recently stumbled upon a fascinating debate in the field of Machine Learning (ML) - the power of 'mind' versus 'data'. While some experts argue that the key to unlocking the full potential of ML lies in architectural designs and inductive biases, others believe that data is the driving force behind its success. 🗣️ Here are some thought-provoking quotes from the experts themselves: 🔹 Dr. Judea Pearl, a Turing Award winner, emphasizes the importance of the 'mind' in his book 'The Book of Why', stating: "Data is profoundly dumb." In a controversial tweet, he also warned that ML people who continue to follow the current data-centric paradigm might be out of a job in three to five years. 🔹 Professor Christopher Manning, director of the Stanford Artificial Intelligence Laboratory, presents a milder opinion, arguing that huge computation and a massive amount of data with a simple learning algorithm create incredibly bad learners. He believes that the structure of the 'mind' allows us to design systems that can learn more from less data. 🔹 On the other side of the debate, Professor Richard Sutton claims that general methods that leverage computation are ultimately the most effective. He argues that researchers who chose to pursue intelligent designs over methods that leverage computation will eventually learn a 'bitter lesson'. 🔹 Peter Norvig, Google’s director of search quality, emphasizes the importance of having a large amount of data over intelligent algorithms in their success: “We don’t have better algorithms. We just have more data.” 💡 My thoughts: I believe that both 'mind' and 'data' are essential for the advancement of ML. However, finding the right balance between the two is crucial. While intelligent algorithms can make better use of limited data, having a large and diverse dataset can help uncover patterns and insights that might not be apparent with smaller datasets. 📢 Now, I'd like to hear your thoughts! Which side of the debate do you lean towards, and why? Do you think there's an ideal balance between 'mind' and 'data' in ML, or will one eventually outweigh the other? 📚 Reference: This discussion was inspired by the book "Designing Machine Learning Systems" by Chip Huyen. For those interested in a deeper dive into the topic, you can find the book here: https://lnkd.in/gj7TW_sn

4

Shashank Sonkar and colleagues are back with another great paper and matching GitHub repo, "Pedagogical Alignment of Large Language Models". You can read the paper on arXiv here - https://lnkd.in/enZfCJgi, and browse the code here - https://lnkd.in/gyW5hSzp. They demonstrate an interesting approach to generating synthetic data to train open weights models to behave more pedagogically, as well as developing some new evaluation metrics. I think there's lots here for those of us interested in this kind of work to learn from and build on. (If synthetic data generation is new do you, check out this post from answer.ai earlier this week - https://lnkd.in/ecsG7R4b.)

38

4 Comments

#RecSys: Interesting work on Tackling the Cold Start Problem in Recommendation Systems. - Anima Singh Nikhil Mehta Trung Vu Xinyang Yi's paper on Better Generalization with Semantic IDs introduces the use of Semantic IDs instead of traditional random IDs for videos, items, and users. Unlike random IDs, which are unconnected by design, Semantic IDs are based on the semantic representation (embeddings) of an item's metadata (e.g., video metadata). This approach leads to better generalization, as similar items/videos are intrinsically linked, significantly improving cold start scenarios. - Jianling Wang Haokai Lu's paper on LLMs for User Interest Exploration showcases a hybrid approach combining LLMs and traditional collaborative filtering. LLMs are used for controlled generation of user interest clusters, leveraging their strength in reasoning and generalization, while collaborative filtering ensures strong personalization. By pre-computing these interest clusters, the approach offers low-latency recommendations and is especially useful for handling cold start cases. Live experiments have shown promising results. It is good example of how LLMs and traditional recommendation techniques are being used together in real-world use cases! Ed H. Chi Lexi Baugher Roger Cheng He Ma Panchapagesan Krishnamurthy Vijay Krishnan Sudarshan Sivaraman Phil Walsh #Turing

42

2 Comments