📝 Weak-Pde-Learn: A Weak Form Based Approach to Discovering PDEs From Noisy, Limited Data 🧠

"Weak-PDE-LEARN uses an adaptive loss function based on weak forms to train a neural network to approximate the PDE solution while simultaneously identifying the governing PDE (see Figure )." [gal30b+] 🤖 #LG

⚙️ https://github.com/punkduckable/Weak_PDE_LEARN
🔗 https://arxiv.org/abs/2309.04699v1 #arxiv

📝 Redundancy-Free Self-Supervised Relational Learning for Graph Clustering 🧠

"A novel self-supervised deep graph clustering method named Relational Redundancy-Free Graph Clustering (R$^2$FGC) is proposed to tackle the problem." [gal30b+] 🤖 #LG

⚙️ https://github.com/yisiyu95/R2FGC
🔗 https://arxiv.org/abs/2309.04694v1 #arxiv

📝 Towards Understanding Neural Collapse: The Effects of Batch Normalization and Weight Decay 🧠

"Provides theoretical guarantees and empirical evidence that neural networks with batch normalization and a high weight decay will exhibit Neural Collapse, whereas neural networks without batch normalization or low weight decay will not." [gal30b+] 🤖 #LG

🔗 https://arxiv.org/abs/2309.04644v1 #arxiv

📝 When Less Is More: Investigating Data Pruning for Pretraining LLMs at Scale 📚🧠

"Performs a rigorous comparison at scale of the simple data quality estimator of perplexity, as well as more sophisticated and computationally intensive estimates of the Error L2-Norm and memorization." [gal30b+] 🤖 #CL #LG

🔗 https://arxiv.org/abs/2309.04564v1 #arxiv

📝 Mobile v-MoEs: Scaling Down Vision Transformers via Sparse Mixture-of-Experts 🔭🧠

"Proposes a simplified and mobile-friendly MoE design where entire images rather than individual patches are routed to the experts to achieve better accuracy and efficiency trade-off on vision tasks." [gal30b+] 🤖 #CV #LG

🔗 https://arxiv.org/abs/2309.04354v1 #arxiv

📝 Learning From Power Signals: An Automated Approach to Electrical Disturbance Identification Within a Power Transmission System 🧠

"Power disturbance events are recorded as a voltage/current waveform over a time period ranging from a few milliseconds to several minutes depending on the event type and the type of recording device." [gal30b+] 🤖 #LG

🔗 https://arxiv.org/abs/2309.04361v1 #arxiv

📝 Encoding Multi-Domain Scientific Papers by Ensembling Multiple CLS Tokens 📚🧠

"Proposes Multi2SPE -- it encourages each of multiple CLS tokens to learn diverse ways of aggregating token embeddings, then sums them up together to create a single vector representation." [gal30b+] 🤖 #CL #DL #LG

🔗 https://arxiv.org/abs/2309.04333v1 #arxiv

📝 Generating the Ground Truth: Synthetic Data for Label Noise Research 🧠

"SYNLABEL generates datasets with a known ground truth function and a soft label distribution, which can be used for label noise injection and measurement of noise-handling methods." [gal30b+] 🤖 #LG

⚙️ https://github.com/sjoerd-de-vries/SYNLABEL
🔗 https://arxiv.org/abs/2309.04318v1 #arxiv

📝 Viewing the Process of Generating Counterfactuals as a Source of Knowledge -- Application to the Naive Bayes Classifier 🧠

"Proposed in this article is based on the fact that, when a counterfactual example is generated, it is also possible to calculate the contribution of each attribute value to the decision of the algorithm [1,2]." [gal30b+] 🤖 #LG

🔗 https://arxiv.org/abs/2309.04284v1 #arxiv

📝 SRN-SZ: Deep Leaning-Based Scientific Error-Bounded Lossy Compression with Super-Resolution Neural Networks 🧠

"By using super-resolution techniques, the proposed SRN-SZ can effectively compress the hard-to-compress scientific datasets, achieving up to 75% compression ratio improvements under the same error bound and up to 80% compression ratio improvements under the same PSNR than the second-best compressor." [gal30b+] 🤖 #LG #DC #IT

🔗 https://arxiv.org/abs/2309.04037v1 #arxiv

📝 Generalization Bounds: Perspectives From Information Theory and PAC-Bayes 🧠👾

"This monograph provides an introduction to information-theoretic generalization bounds, and their connection to the PAC-Bayesian framework, which provides a general framework for studying the generalization capabilities of machine learning algorithms." [gal30b+] 🤖 #LG #AI #IT

🔗 https://arxiv.org/abs/2309.04381v1 #arxiv

📝 Optimal Transport with Tempered Exponential Measures 🧠

"Generalizes Sinkhorn algorithm to $\mathcal{F}_{\alpha}$, a new class of cost matrices which includes the classical cost as a special case." [gal30b+] 🤖 #LG

🔗 https://arxiv.org/abs/2309.04015v1 #arxiv

📝 Multimodal Transformer for Material Segmentation 🔭🧠

"Proposes a fusion strategy that can effectively fuse information from different combinations of multiple modalities including RGB, Angle of Linear Polarization (AoLP), Degree of Linear Polarization (DoLP) and Near-Infrared (NIR)." [gal30b+] 🤖 #CV #LG

⚙️ https://github.com/csiplab/MMSFormer
🔗 https://arxiv.org/abs/2309.04001v1 #arxiv

📝 Large-Scale Automatic Audiobook Creation 🔊👾🧠

"Leverages recent advances in neural text-to-speech and text summarization and allows users to customize an audiobook's speaking style and speed using a small amount of speech samples." [gal30b+] 🤖 #SD #AI #DC #DL #LG

🔗 https://arxiv.org/abs/2309.03926v1 #arxiv

📝 Active Learning for Classifying 2D Grid-Based Level Completability 🧠👾

"Uses active learning to query levels to label with completability and train deep-learning models to classify the completability of generated levels for Super Mario Bros, Kid Icarus, and a Zelda-like game." [gal30b+] 🤖 #LG #AI

⚙️ https://github.com/MahsaBazzaz/level-completabilty-x-active-learning
🔗 https://arxiv.org/abs/2309.04367v1 #arxiv

📝 DBsurf: A Discrepancy Based Method for Discrete Stochastic Gradient Estimation 🧠

"Introduces DBsurf, an estimator for discrete distributions that uses a novel sampling procedure to reduce the discrepancy between the samples and the actual distribution, thereby improving gradient estimation." [gal30b+] 🤖 #LG

🔗 https://arxiv.org/abs/2309.03974v1 #arxiv

📝 UER: A Heuristic Bias Addressing Approach for Online Continual Learning 🧠🔭

"UER learns current samples only by the angle factor and further replays previous samples by both the norm and angle factors to address the bias problem in continual learning, achieving superior performance over various state-of-the-art methods." [gal30b+] 🤖 #LG #CV

⚙️ https://github.com/FelixHuiweiLin/UER
🔗 https://arxiv.org/abs/2309.04081v1 #arxiv

📝 Sample-Efficient Co-Design of Robotic Agents Using Multi-Fidelity Training on Universal Policy Network 🦾🧠

"Proposes to use Hyperband as a multi-fidelity optimization strategy to improve efficiency of the Co-design optimization by warm starting the control optimization using a universal policy learner that ties the controllers learnt across the design spaces." [gal30b+] 🤖 #RO #LG

🔗 https://arxiv.org/abs/2309.04085v1 #arxiv

📝 ConDA: Contrastive Domain Adaptation for AI-generated Text Detection 📚👾🧠

"Develops a contrastive domain adaptation framework, called ConDA, which learns domain-invariant feature representations via a contrastive loss in conjunction with standard domain adaptation techniques such as DANN and CDAN." [gal30b+] 🤖 #CL #AI #LG

⚙️ https://github.com/AmritaBh/ConDA-gen-text-detection
🔗 https://arxiv.org/abs/2309.03992v1 #arxiv

📝 Improving Resnet-9 Generalization Trained on Small Datasets 🧠🔭

"A combination of various techniques to improve generalization including sharpness aware optimization, label smoothing, gradient centralization, input patch whitening as well as metalearning based training." [gal30b+] 🤖 #LG #CV

🔗 https://arxiv.org/abs/2309.03965v1 #arxiv