Anahita Pakiman

AnchorGT: A Novel Attention Architecture for Graph Transformers as a Flexible Building Block to Improve the Scalability of a Wide Range of Graph Transformer Models Quick read: https://lnkd.in/gFiQyv8a Paper: https://lnkd.in/gsJqpi5F #machinelearning #ai

18

🚀 Exciting News! 🚀 We are thrilled to announce that Kumo AI's position paper "Relational Deep Learning: Graph Representation Learning on Relational Databases," has been accepted at #ICML 2024! 🎉 Join us in Vienna from July 21-27, where our very own Matthias Fey will be presenting this exciting new research area. 🔗 https://icml.cc/ In collaboration with Stanford University, we introduce Relational Deep Learning (#RDL), a new research area for applying deep learning on relational databases by viewing them as temporal, heterogeneous graphs. This method eliminates the need for manual feature engineering, leading to more accurate models built faster. 🚀 We’ve also developed #RelBench, a benchmark suite to facilitate research in this new area, using diverse datasets from Stack Exchange discussions to Amazon book reviews. 📚 Stay tuned for more updates on our presentation format, and check out our paper here: https://lnkd.in/dk24vbSg #ICML2024 #MachineLearning #AI #GraphNeuralNetworks #RelationalDeepLearning #DataScience #KumoAI #StanfordCollaboration

38

In Daniel Klitzke's latest article, he shows how you can use the Huggingface Transformers and Sentence Transformers libraries to boost you RAG pipelines using reranking models. #LLM #MachineLearning

22

Tomorrow’s the day! #Nodes2024 by Neo4j is almost here! Join Dr. Benjamin Niedermann on November 7, 1:00 PM - 1:30 PM UTC for an insightful talk on “Advanced #GraphVisualizations in #Jupyter Notebooks. Register here: https://lnkd.in/eudwj68K #Nodes2024 #Neo4j #GraphTechnology #GraphDatabase #DataVisualization #GraphVisualization #JupyterNotebooks #DataScience #AdvancedAnalytics #DataInnovation #KnowledgeGraph #DataInsights #GraphDataScience 

7

Interested in a powerful and efficient Transformer model for graph data? Welcome to check out our upcoming paper at ICLR'24, titled “Polynormer: A Polynomial-Expressive Graph Transformer in Linear Time”. The concept of learning polynomial functions dates back to the last century. Our work revisits this idea by integrating it into the Transformer model. We theoretically demonstrate Polynormer’s expressivity by showcasing its ability to learn high-degree polynomials in an efficient manner. Our experimental results demonstrate that Polynormer achieves state-of-the-art results across various important graph datasets, including Google TpuGraphs, an industrial-scale graph benchmark used for predicting AI model run time on TPUv3. 📜 Paper: https://lnkd.in/eBh77e8d 👨‍💻 Code: https://lnkd.in/e9-nFxHx 📺 Talk: A 5-min presentation by Chenhui Deng will be hosted on ICLR'24 website and our YouTube channel (https://lnkd.in/eK4Y6ShT) for asynchronous viewing after the conference.

38

Discover potential paradigm changes for data and analytics💡 BISE Journal recently published a new article by Ivo Blohm, Felix Wortmann, Christine Legner, and our own Dr. Felix Köbler. 🔎Check out the article "Data products, data mesh, and data fabric" which explores these concepts, highlighting their potential to revolutionize data management and analytics: https://lnkd.in/e7F-z6p6 #DataProducts #DataMesh #DataFabric #DataProductManagement

11

🔎 𝗫𝗔𝗜 𝗖𝗼𝗻𝗳𝗲𝗿𝗲𝗻𝗰𝗲 𝟮𝟬𝟮𝟱 - 𝗦𝗽𝗲𝗰𝗶𝗮𝗹 𝗧𝗿𝗮𝗰𝗸 𝗦𝗽𝗼𝘁𝗹𝗶𝗴𝗵𝘁: Debugging and improving models with xai 🖥️ ✏️ 𝗔𝘂𝘁𝗵𝗼𝗿𝘀: @christopher j anders Sebastian Lapuschkin Frederik Pahde Maximilian Dreyer Lorenz Linhardt 🔗 𝗟𝗲𝗮𝗿𝗻 𝗠𝗼𝗿𝗲 𝗛𝗲𝗿𝗲: https://lnkd.in/dz2UZf9T This track aims to debug, modify, and evaluate machine learning models to enhance their performance and reliability using eXplainable AI (XAI) techniques. The topics include, but are not limited to, (1) identifying and mitigating the effect of learned spurious correlations (also known as Clever Hans effect or shortcut learning) and distribution shifts in the data, (2) diagnosing failure modes and edge cases leading to undesired model behaviour, and (3) improving models through targeted corrections. For each of these points, XAI methods serve as a critical foundation, as they enable to understand model decision strategies (and their flaws), and to identify critical parts of the model or data set that are most amenable to improvement. Additionally, this track covers the development and application of rigorous evaluation methods for verifying that the enhancements achieve the intended outcomes. Join us to discuss best practices, emerging tools, innovative techniques, and case studies showcasing how thoughtful model refinement can lead to fair, robust, and responsible AI solutions. #XAI #debug #machinelearning #spurious #correlations #cleverhans #understanding #model #decisions #artificialintelligence

56

Preprint analysis focusing on the use of AI as a tool in #OpenScience. Thinking about these AI-related challenges and opportunities at your organization? Delta Think can help you understand, define, and implement your next steps. Please reach out today to get started. #ScholComm #AI

2

How to Selectively Scale Numerical Input Variables for Machine Learning https://zurl.co/prqb

33

Learn Parameter Tuning in #DeepLearning! Unlock the secrets to fine-tuning your deep learning models with our cheat sheet on parameter tuning. Learn how to initialize weights, leverage transfer learning, and optimize convergence for better performance. 👉 𝐓𝐨𝐩𝐢𝐜𝐬 𝐂𝐨𝐯𝐞𝐫𝐞𝐝: 👉 𝐖𝐞𝐢𝐠𝐡𝐭𝐬 𝐈𝐧𝐢𝐭𝐢𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 🔹 𝐗𝐚𝐯𝐢𝐞𝐫 𝐈𝐧𝐢𝐭𝐢𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 Discover how this method helps in setting initial weights that consider the unique architecture of your model, leading to more stable training. 🔹 𝐓𝐫𝐚𝐧𝐬𝐟𝐞𝐫 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 Learn how to save time and resources by using pre-trained weights from large datasets, applying them to your specific tasks. 👉 𝐎𝐩𝐭𝐢𝐦𝐢𝐳𝐢𝐧𝐠 𝐂𝐨𝐧𝐯𝐞𝐫𝐠𝐞𝐧𝐜𝐞 🔹 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐑𝐚𝐭𝐞 Understand the importance of the learning rate and how it influences the pace at which your model learns. 🔹 𝐀𝐝𝐚𝐩𝐭𝐢𝐯𝐞 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐑𝐚𝐭𝐞𝐬 Explore how varying the learning rate can speed up training and lead to better results, with techniques like the Adam optimizer. 📚These insights will help you fine-tune your models more effectively, leading to faster training and improved accuracy. 👉 𝐈𝐦𝐩𝐨𝐫𝐭𝐚𝐧𝐜𝐞 𝐨𝐟 𝐋𝐞𝐚𝐫𝐧𝐢𝐧𝐠 𝐓𝐡𝐞𝐬𝐞 𝐂𝐨𝐧𝐜𝐞𝐩𝐭𝐬: Understanding parameter tuning is crucial for anyone serious about deep learning. Proper weight initialization, smart use of transfer learning, and optimizing convergence are key to building models that learn efficiently and perform exceptionally well. 📌 Follow 1stepGrow Academy for more insights on data science, machine learning & Ai. 📍 Passionate about data science and AI? Turn your passion into expertise at 👉 https://lnkd.in/dZte5GdH #1stepGrow #ParameterTuning #AI #MachineLearning #DataScience

1

Introducing a new approach to graph-enabled RAG. LazyGraphRAG needs no prior summarization of source data, avoiding prohibitive up-front indexing costs. It’s inherently scalable in cost and quality across multiple methods and search mechanisms: https://lnkd.in/eMsfeiey

513

12 Kommentare

Dive into video segmentation with AI! This article offers a detailed guide on using SAM for effective segmentation. A valuable resource for Data Scientists exploring solutions in this area. #SAM #computervision #deeplearning #videosegmentation #aiml

4

In the realm of ML, particularly DL with artificial neural networks, two concepts reign supreme: ⚫️Gradient Descent and the Loss Function. These two, though seemingly separate, are intricately intertwined, working together to guide models towards optimal performance. ⚫️Gradient Descent is an optimization algorithm, a powerful tool used to achieve the objective of minimizing the Loss Function. It works by iteratively adjusting the internal parameters of the model, such as weights and biases, to steer the model towards the direction of lower loss. And this is the ultimate goal in training a model – to minimize this Loss Function. # This code defines a LinearRegression class that implements a basic linear regression model: import numpy as np class LinearRegression:   def __init__(self, learning_rate=0.01):     self.learning_rate = learning_rate     self.weights = None     self.bias = None # The fit function trains the model on the provided data (X, y).   def fit(self, X, y):     # Initialize weights and bias with zeros     self.weights = np.zeros(X.shape[1])     self.bias = 0 # Training loop - we calculate the predictions (y_predicted) using the current weights and bias:     for _ in range(1000): # Forward pass - calculate predictions       y_predicted = X.dot(self.weights) + self.bias # Mean Squared Error (MSE) loss function, measuring the average squared difference between predictions and actual values:       loss = np.mean((y - y_predicted) ** 2) # Calculate gradients of loss w.r.t. weights and bias       weight_gradient = -2 * X.T.dot(y - y_predicted) / X.shape[0]       bias_gradient = -2 * np.mean(y - y_predicted) # Update weights and bias using gradient descent       self.weights -= self.learning_rate * weight_gradient       self.bias -= self.learning_rate * bias_gradient   def predict(self, X):     return X.dot(self.weights) + self.bias ⚫️The relationship between Gradient Descent and the Loss Function is a beautiful example of collaboration in Machine Learning. The Loss Function acts as the objective measure, guiding the model in the right direction, while Gradient Descent serves as the optimizer, making the necessary adjustments to reach the goal. By working together, they enable us to train powerful models that can learn from data and make accurate predictions. #ML #DL #ANNs #GradientDescent #LossFunction #WeightsAndBiases #MetastabilityInc #ITsolutions #WebDevelopment #SoftwareDevelopment #Business #AI #FutureofAI #ITsolutions #BusinessDevelopment #CyberSecurity #AppDevelopment #MobileAppDevelopment #CustomSoftware #Devops #API #DataScience #Algorithms

3

🔍 Ever wondered how experiments with small-scale models translate to real-world applications? We often get asked, "How do you compare model size experiments with real-life scenarios?" and even, "Why are there LEGO-men on top of your models?" While it might seem complex to scale down both size and time for an experiment, the principles behind it are incredibly fascinating. 🎥 In this video, Morten Kramer takes us behind the scenes at Aalborg University's state-of-the-art basin to explain the magic of Froude scaling. He'll show you how our small-scale models, complete with LEGO figures, can accurately predict real-world outcomes. It's all about turning miniature experiments into full-scale data, bridging the gap between theory and practical application. Don't miss this chance to gain insight into how scaling in model testing is not just possible but essential for accurate real-world simulations. Whether you're a seasoned engineer or simply curious about how science brings models to life, this video is for you! 🔗 Watch the video and join the conversation! #EngineeringExcellence #FroudeScaling #ModelTesting #RealWorldApplications #Innovation #AalborgUniversity #Hydrodynamics #ScaleModeling #STEM #ResearchAndDevelopment

39