results.csv

Paper ID,Publishable,Conference,Rationale
P001,0,na,na
P002,0,na,na
P003,0,na,na
P004,1,TMLR,"The proposed research on Training-Free Graph Neural Networks (TFGNNs) utilizing labels as features (LaF) addresses the computational challenges of Graph Neural Networks (GNNs), providing a practical solution for real-world scenarios with limited resources. The paper demonstrates improved performance in transductive node classification tasks compared to existing methods, making it a valuable contribution to the field of machine learning research, particularly suitable for the Transactions on Machine Learning Research (TMLR) conference. 
P005,1,CVPR,"The paper presents an unsupervised method for clothing recognition and retrieval, introducing a new dataset of high-resolution street fashion photos with detailed annotations. The proposed method's effectiveness is demonstrated, and it's scalable for large-scale applications. Its potential impact on the online clothing sales industry makes it a suitable submission for the Computer Vision and Pattern Recognition (CVPR) conference."
P006,0,na,na
P007,1,EMNLP,"The proposed research paper is suitable for the EMNLP conference, as it introduces an innovative end-to-end discourse parser that jointly parses at both syntax and discourse levels, a first in the field. It also creates the first syntacto-discourse treebank, integrating the Penn Treebank with the RST Treebank, and requires no preprocessing, making it self-contained and efficient. These significant contributions and focus on end-to-end models align well with the EMNLP conference's themes."
P008,1,KDD,"The proposed research, utilizing machine learning models and Chain-of-Thought (CoT) prompting, investigates the effects on large language models' sample complexity, approximation power, and generalization. It offers theoretical and empirical evidence, contributing to AI by deepening understanding of CoT principles, potentially leading to improved LLM training methods. The research's relevance lies in its data-driven analysis, complex task focus, and exploration of sample complexity, making it suitable for the KDD conference."
P009,0,na,na
P010,1,NEURIPS,"This paper, ""Model-based Counterfactual Advantage Learning for Sample-Efficient Sequential Recommendation,"" introduces MBCAL, a reinforcement learning method that reduces learning variance through a masked environment model and future advantage model. The MBCAL method outperforms existing approaches in both sample efficiency and asymptotic performance. The paper's focus on sequential decision-making, sample efficiency, and variance reduction makes it a suitable submission for the NeurIPS conference, aligning with its focus on machine learning, artificial intelligence, and optimization."
P011,1,NEURIPS,"The research paper is suitable for the AI and Statistics track at NEURIPS as it offers innovative statistical methods (HTE-BH and HTE-Knockoff) for detecting HTE in A/B testing scenarios, controlling FDR, and addressing scalability issues. It provides theoretical and practical evidence using data from Snap, aligning with the conference's focus on large-scale machine learning problems and the development of statistical methods in AI."
P012,0,na,na
P013,0,na,na
P014,0,na,na
P015,0,na,na
P016,0,na,na
P017,1,KDD,"The proposed research targets video analysis highlight detection and summarization, employing crowdsourced time-synchronized comments and natural language processing. It introduces an unsupervised framework, utilizing lag-calibration via concept-mapped lexical chains and a modified SumBasic algorithm. The research presents innovative methods, creates extensive datasets, and develops an emotion lexicon for bullet-comments, all in line with the International Conference on Knowledge Discovery and Data Mining's focus on data mining and knowledge discovery."
P018,1,NEURIPS,"The research ""Plasticity Injection"" aims to tackle plasticity loss in deep reinforcement learning, a persistent issue causing neural networks to lose learning effectiveness. It proposes a novel solution, ""plasticity injection,"" based on proactive diagnosis, targeted mitigation, and dynamic capacity expansion. Demonstrating improvements in long-term performance and learning stability, the research showcases significant advancements on challenging RL benchmarks, including continuous control tasks and partially observable environments. Additionally, the diagnostic component provides insights into plasticity loss mechanisms, deepening our understanding of this critical RL issue."
P019,1,KDD,"The paper proposes a novel framework for analyzing TB care data from Digital Adherence Technologies (DATs), focusing on unrecorded interventions. It demonstrates the effectiveness of this approach in real-time adherence, outcome prediction, and identifying high-risk patients. The methodologies could be applied to any medication schedule, making it relevant to the KDD community. Key contributions include a proxy for interventions, a model for prioritizing intervention targets, and adapting the LEAP model for specific applications."
P020,0,na,na
P021,0,na,na
P022,0,na,na
P023,1,NEURIPS,"The paper proposes a novel reverse hierarchy model for predicting eye fixations, based on the Reverse Hierarchy Theory (RHT), using stochastic fixation models and image super-resolution techniques. It applies concepts like compressive sensing and Gaussian maps, and discusses potential extensions in computer vision, aligning with the interdisciplinary focus of NeurIPS. Keywords include layer, fixation, detection, xh, eye, images, information, compressive, attention, ak1, maps, sensing, adopted, super, proposed."
P024,1,NEURIPS,"The presented research paper is suitable for NEURIPS, focusing on machine learning, specifically attack and defense mechanisms, robustness, and vulnerabilities in models. It introduces a novel label-only backdoor attack, FLIP, testing its effectiveness across various scenarios, including real-world data challenges. The paper's findings offer insights into model vulnerabilities, contributing to both attackers and defenders. Its focus on the attack's efficiency, robustness against defense mechanisms, and generalizability across datasets and model architectures aligns with NEURIPS' interdisciplinary approach, bridging theory and practice in machine learning. The paper also explores the CTA-PTA trade-off in the security of knowledge"
P025,1,CVPR,"The novel scene parsing algorithm in this research enhances per-pixel and per-class accuracy by combining likelihood scores from multiple classification models and incorporating global context. It addresses low recognition rates of foreground classes, produces semantically relevant parsing output, and delivers impressive results on the SIFTflow dataset, with comparable performance to state-of-the-art methods on the LMSun dataset. The method's use of a balanced dataset, comparison with baseline methods, and in-depth analysis make it a valuable addition to the CVPR conference."
P026,0,na,na
P027,1,KDD,"The KDD conference is suitable for the research paper ""emoji2vec: Learning Emoji Embeddings from Unicode Descriptions"" as it focuses on natural language processing and data mining applications in social media. The paper introduces a method for generating emoji embeddings, outperforming the skip-gram model in a Twitter sentiment analysis task. It also discusses future improvements for context-dependent emoji definitions, such as sarcasm. The approach works on Unicode descriptions, enabling future Unicode symbol embeddings."
P028,1,EMNLP,"The proposed research paper focuses on developing a novel framework to understand ambiguous language in a visual context, addressing a fundamental task for humans and a key challenge for language acquisition. It introduces a new task for resolving structural ambiguities using visual signals and presents an extension to the sentence tracker approach, enabling multiple interpretations of a sentence based on video compatibility. The paper utilizes a multimodal corpus, LAVA, containing sentences with linguistic ambiguities paired with short videos, to study ambiguity-related phenomena in visually grounded language processing."
P029,1,KDD,"The research paper ""OpenOmni"" presents an open-source framework for multimodal conversational agents, addressing real-world challenges in balancing cost, latency, and accuracy. It emphasizes the importance of Retrieval-Augmented Generation (RAG) and highlights potential applications, such as indoor assistance for the visually impaired. The KDD conference, focusing on data science, machine learning, and AI, offers an ideal platform for disseminating this innovative framework."
P030,1,NEURIPS,"The research paper proposes a novel approach for optimizing memory in dynamic shape graph tasks in deep learning, using symbolic shapes for operation scheduling and rematerialization. It demonstrates significant memory reduction during training and aligns with NeurIPS' focus on advancing neural information processing systems, potentially impacting the compiler community and increasing the use of dynamic shape compilers. The paper's use of exact symbolic shapes, runtime analysis, and potential for improving computation efficiency make it a valuable contribution to the conference."
P031,0,na,na
P032,0,na,na
P033,0,na,na
P034,1,NEURIPS,"The research paper ""Evaluating Dual PatchNorm"" is suitable for NeurIPS due to its exploration of Layer Normalization's impact on Vision Transformers, aligning with the conference's focus on machine learning, neural networks, and computer vision. The innovative, rigorous methodology showcases the authors' approach to evaluating various LayerNorm placements. Notable findings include significant performance improvement with Dual PatchNorm on natural datasets, contributing to ViTs optimization. The paper's extensive experiments across datasets and learning rates demonstrate broad applicability and reproducibility."
P035,1,KDD,"This research paper proposes the use of unconventional algorithms, inspired by slime molds, to optimize complex crowdsourced delivery networks. It employs game-theoretic optimization and evolutionary game theory to model strategic decision-making processes, focusing on cooperative behavior and potential ""cheating"" strategies. Aligning with KDD's focus on innovative data mining and knowledge discovery, the paper's application of these techniques is particularly relevant."
P036,0,na,na
P037,1,EMNLP,"The research paper proposes a Chinese span-extraction dataset for machine reading comprehension, featuring 20,000 questions annotated by experts. It reports impressive results on development and test sets, yet models struggle with the challenge set, which involves complex reasoning. The paper highlights the dataset's potential for boosting linguistic diversity in the field and encouraging further research on comprehensive reasoning questions, using empirical methods like annotated datasets and evaluations."
P038,0,na,na
P039,0,na,na
P040,0,na,na
P041,0,na,na
P042,0,na,na
P043,0,na,na
P044,1,NEURIPS,"The research paper aligns with NeurIPS, focusing on the application of DNNs to agricultural planning, specifically for county-level crop yield predictions in the U.S. It introduces a novel, multi-modal dataset, CropNet, which combines Sentinel-2 Imagery, WRF-HRRR Computed Dataset, and USDA Crop Dataset for high-resolution imagery, weather conditions, and yield data. The paper aims to enhance prediction accuracy by considering weather variations and climate change impacts, and discusses the potential of CropNet to improve DNNs' generalization capabilities."
P045,0,na,na
P046,1,TMLR,"The proposed paper on adversarial robustness in graph neural networks, particularly the symbiotic attack model, offers a novel, comprehensive threat model and a scalable, memory-efficient solution. Its focus on practical applications and potential for future research aligns with the TMLR conference's emphasis on machine learning and robotics innovations."
P047,0,na,na
P048,0,na,na
P049,1,CVPR,"The proposed research paper showcases an innovative self-adaptation method for enhancing semantic segmentation models' robustness against out-of-distribution data. This adaptive inference process improves model predictions without modifying the training process or architecture. The method outperforms the baseline in multiple applications, including cityscapes, and offers a beneficial trade-off between accuracy and computational cost compared to model ensembles. The paper follows best practices in machine learning research, outlines an evaluation process, and discusses potential future research directions, making it a suitable fit for the CVPR conference."
P050,1,KDD,"The research paper aligns with KDD's focus on data mining and knowledge discovery by exploring new strategies for interpreting deep learning models in the context of Natural Language Inference (NLI). It offers insights into deep learning model behavior, extends to other NLP tasks, and uses relevant keywords like ""learning,"" ""data,"" ""mining,"" ""interpretation,"" and ""visualization."""
P051,0,na,na
P052,0,na,na
P053,0,na,na
P054,1,CVPR,"The proposed research on 3D food reconstruction techniques, including a challenge to push boundaries, aligns with CVPR's focus on computer vision and its applications. It highlights the importance of accurate volume estimation and shape reconstruction in nutritional analysis and food presentation. The paper utilizes innovative mesh reconstruction methods and evaluates performance using Chamfer distance and MAPE, making it relevant to the CVPR community."
P055,1,NEURIPS,"The proposed research paper is suitable for NeurIPS as it investigates the effects of a new requirement for broader impact statements in technical research communities, aligning with NeurIPS's mission for socially beneficial AI research. The paper's findings on researchers' perspectives and processes in drafting these statements, as well as its emphasis on ethical AI practices, are relevant to NeurIPS's commitment to responsible AI. Key findings, like the division in the requirement's framing and the need for ongoing monitoring, can contribute to broader AI community discussions on ethical practices."
P056,0,na,na
P057,0,na,na
P058,1,TMLR,"The proposed research, focusing on positional encoding in Recurrent Neural Networks (RNNs), presents an innovative approach to enhancing RNNs' capabilities, particularly in handling diverse input sequences and larger vocabularies. It challenges the conventional view of positional encoding as a substitute for RNNs' temporal processing. The study, based on synthetic benchmarks, offers insights into the potential benefits of positional encoding for RNNs, aligning with TMLR's emphasis on model training strategies, making it a relevant and engaging submission for the conference."
P059,1,EMNLP,"The research paper proposes an innovative method of enhancing Recurrent Neural Networks (RNNs) performance by incorporating positional encoding, traditionally used in Transformers. This approach enables RNNs to manage larger vocabularies and a broader range of discrete inputs. The paper's use of synthetic benchmarks, exploration of positional encoding's impact on RNNs, and discussion of time perception align with the EMNLP conference's focus on natural language processing, empirical methods, and interdisciplinary research."
P060,1,TMLR,"The proposed research paper is suitable for the TMLR conference as it focuses on developing machine learning algorithms for object classification and segmentation in video sequences. It presents a novel method using location-specific variances in background and foreground models to enhance accuracy. The approach optimizes kernel variance for classification and object properties, mitigates erroneous classification, and aligns with the conference's focus on learning and adapting to data properties. The paper's clear examples and visualizations make it a great fit for TMLR."
P061,1,TMLR,"The paper offers an innovative contrastive instance discrimination approach for self-supervised learning, addressing the loss of semantic features due to random cropping. It proposes a solution by including the original image during training, demonstrating effectiveness across various benchmark datasets and outperforming several SOTA instance discrimination SSL methods. The paper's focus on performance, use of contrastive learning, and its impact on SSL make it suitable for the TMLR conference, emphasizing machine learning research and applications."
P062,1,NEURIPS,"The research paper proposes a novel method for adapting large pretrained models while preserving equivariance, crucial for symmetric domains. It uses regularization techniques to maintain equivariance during adaptation. The paper's significant performance improvements over existing techniques, along with its alignment with NeurIPS's focus on new machine learning algorithms, makes it an ideal fit for the NeurIPS conference."
P063,1,NEURIPS,"The research paper, ""Two Parallel AI Communities: A Study of Cultural Influence on Research Priorities,"" is suitable for NeurIPS. It explores cultural differences in AI research, particularly between American and Chinese researchers, aligning with NeurIPS's mission. The paper's insights into differing research interests, impacts on AI evolution, and analysis of architectures like PCANet and Deep Forests offer unique perspectives. It also discusses ethical AI, lack of collaboration in fairness, and potential for knowledge transfer, resonating with NeurIPS's commitments."
P064,1,NEURIPS,"The proposed research paper is suitable for NeurIPS as it focuses on enhancing open-source large language models' multi-modal tool handling for visual comprehension and image generation. It introduces a self-instruction framework that enables models to utilize a wide variety of tools, filling a gap compared to proprietary models. The paper's emphasis on simplicity, efficiency, and modular design aligns with NeurIPS' focus on practical machine learning applications. It employs prompt engineering, reinforcement learning, and demonstrates significant performance improvements across visual tasks, showcasing its theoretical and practical contributions. Keywords include capabilities, range, and integration."
P065,1,TMLR,"The research paper proposes a study on recursive inpainting in generative AI, addressing model collapse issues. It empirically investigates the effects of recursively applying AI image models, offering insights into recursion's impact on model performance. These findings, focusing on content distance and loop processes, are crucial for developing more robust AI models, aligning well with the TMLR conference's focus on machine learning research."
P066,1,TMLR,"This research paper proposes a novel method for compressing language model vocabulary through domain-specific tokenizer training, aiming to reduce computational costs without sacrificing performance. The approach aligns with TMLR's focus on efficient NLP techniques, using a teacher-student framework and exploring tokenization in various domains. The paper's emphasis on compact and meaningful inputs resonates with TMLR's interest in practical, scalable solutions for NLP problems."
P067,0,na,na
P068,1,CVPR,"The Feature Flow Net (FFNet) paper is a strong candidate for the CVPR conference, as it presents a novel framework for VIC3D object detection that addresses temporal asynchrony and limited bandwidth issues by transmitting compressed feature flow. The paper's strengths include its ability to capture dynamic scenes, adaptive compression scheme, and robustness to temporal misalignment. The paper is well-structured, clearly written, and provides robust experimental results demonstrating FFNet's superiority over existing methods. Keywords include raw data, optical maps, efficiency, and robustness."
P069,0,na,na
P070,0,na,na
P071,1,EMNLP,"The paper proposes a novel deep learning approach for analyzing fillers, a type of disfluency, in spoken language. It addresses the importance of fillers in spoken language processing and presents a case study on a spontaneous speech corpus, demonstrating their value in predicting confidence and sentiment. The paper's key contributions align with the EMNLP conference's focus on empirical methods, offering original research and experimental results using contextualized embeddings."
P072,0,na,na
P073,0,na,na
P074,0,na,na
P075,1,NEURIPS,"The research paper proposes a novel pipeline for multi-step inference in instructional videos, using VideoCLIP, GRU, and attention mechanisms. It conducts extensive experiments to validate its method, contributes to the AQTC challenge, and aligns with the conference's focus on AI and machine learning, making it a significant contribution to the NEURIPS community."
P076,0,na,na
P077,0,na,na
P078,0,na,na
P079,0,na,na
P080,1,EMNLP,"This research paper investigates the application of large language models (LLMs) in scientific workflows, evaluating their performance in tasks like paper error identification, result comparison, and checklist compliance. Using GPT-4, it presents original research through small-sample experiments, providing qualitative evidence of LLM impact on authors' submissions. It also discusses potential LLM manipulation and its ethical implications, making it a suitable fit for EMNLP due to its focus on natural language processing, experimental approach, and ethical considerations."
P081,0,na,na
P082,0,na,na
P083,0,na,na
P084,0,na,na
P085,1,TMLR,"The paper proposes a technical assessment framework for individual privacy in synthetic data sets, addressing a gap in the literature. It discusses privacy metrics, risks, and safe/unsafe models, connecting to adversaries and predictors. The paper's interdisciplinary focus aligns with TMLR's goal of fostering collaboration in machine learning, cybersecurity, and related fields, making it an excellent fit for the TMLR conference."
P086,0,na,na
P087,1,CVPR,"The research paper proposes a novel video feature tracking method using low-rank regularization in the Lucas-Kanade optimization problem. It performs well in both rigid and non-rigid environments, handling dynamic scenes with multiple moving objects. Experiments with various low-rank regularizers are presented using a unified optimization framework for real-time tracking. Results show improved tracking quality compared to traditional methods, making it a significant contribution to computer vision, particularly relevant to the CVPR conference with its focus on real-time tracking and real-world applications."
P088,1,NEURIPS,"This research paper, ""Quantifying Modularity in Neural Networks"" proposes a practical definition of modules in ANNs, presents methods for assessing functional similarity, and highlights differences between upstream and downstream perspectives in neural representation. The paper's empirical evaluation is based on over 250 feedforward, fully-connected neural networks trained on MNIST. Findings suggest that representation functions may not align between upstream and downstream definitions, a crucial observation for interpreting neural representations. The paper aligns with NeurIPS conference themes focusing on modularity, neural representations, and upstream/downstream perspectives."
P089,1,NEURIPS,"The research paper is suitable for NEURIPS due to its focus on the NTK approximation, a valuable tool for analyzing neural network training dynamics. It presents new results on its validity, applicability, and deviations, bridging the gap between theory and practice. Additionally, it offers new theoretical bounds and rescaling techniques, significant contributions in machine learning and neural networks. The paper's focus on the NTK approximation and its implications for neural network training dynamics make it a fitting submission for NEURIPS."
P090,1,NEURIPS,"The paper presents a novel regularization scheme derived from group representation theory, addressing the limitations of traditional fine-tuning in preserving equivariance during model adaptation. The method, evaluated on diverse benchmark datasets, consistently outperforms traditional fine-tuning and state-of-the-art adaptation techniques. Relevant to NEURIPS, the paper's contributions align with the conference's focus on machine learning, AI, and computational neuroscience, offering insights into improving generalization performance."
P091,1,CVPR,"This paper, focusing on applying BiLSTM-CRF for redacting sensitive data from call center transcripts, is a promising submission for CVPR. It addresses real-world challenges in speech recognition and NLP, including poor audio quality and ungrammatical sentences. The paper's innovative solution could significantly impact data privacy, making it a valuable contribution to the field."
P092,0,na,na
P093,0,na,na
P094,0,na,na
P095,1,KDD,"The paper ""JueWu-MC: A Sample-efficient Hierarchical RL Framework for Playing Minecraft"" is relevant for KDD due to its application of advanced RL techniques in a real-world Minecraft environment, outperforming existing solutions in the MineRL competition. It employs representation and imitation learning, aligning with KDD's focus on data mining and ML. The paper discusses challenges in Minecraft, such as sparse rewards and high-dimensional visual input, indicating potential for further research."
P096,0,na,na
P097,0,na,na
P098,0,na,na
P099,1,EMNLP,"This research paper, focusing on integrating linguistic knowledge into video-captioning models using LSTM RNNs, is an innovative contribution to the NLP community. The authors employ various techniques and conduct experiments on three large video-caption datasets, demonstrating the impact on grammar and descriptive quality. The paper's findings could be applicable to other image and video captioning models, making it suitable for the EMNLP conference."
P100,0,na,na
P101,1,CVPR,"The proposed research applies deep learning and multiple instance learning (MIL) for emphysema detection in CT scans, overcoming traditional MIL limitations by maintaining inter-sample relationships using a novel CNN-based approach. The paper's focus on 3D imaging and pooling strategies in MIL-based models, as well as its alignment with CVPR's emphasis on computer vision and medical imaging, makes it an ideal fit for the CVPR conference."
P102,1,CVPR,"The ""CompCars"" research paper is suitable for the CVPR conference, offering a unique large-scale car image dataset with car hierarchy, attributes, viewpoints, and parts. The paper presents deep learning experiments predicting car attributes and discusses cross-scenario car analysis challenges. Its focus on Bayesian methods and impressive results make it a valuable contribution to the field, aligning with the CVPR conference's research focus."
P103,0,na,na
P104,1,EMNLP,"The research paper ""A Scalable Approach to Enhancing the Self-Consistency of Pre-trained Language Models"" is a suitable submission for EMNLP, as it presents ConCoRD, a novel method using pre-trained NLI models to improve model predictions' consistency, outperforming existing methods on various datasets, such as BeliefBank, Natural Questions, and ConVQA, while also comparing the performance of different NLI models."
P105,0,na,na
P106,0,na,na
P107,0,na,na
P108,1,KDD,"The proposed research on phoneme inventory induction, integrating phonological knowledge into speech technologies for under-resourced languages, aligns with KDD's objectives. It uses cross-linguistic consistency analysis and has potential for improving multilingual NLP, speech technology, and language documentation, making it relevant to KDD's emphasis on data-driven solutions for real-world problems."
P109,1,CVPR,"The paper proposes a novel method to detect hate speech in multimodal memes, leveraging Transformer models and a bidirectional cross-attention mechanism. It significantly outperforms baselines and enhances multimodal understanding, making it a valuable contribution to the CVPR conference's focus on computer vision and machine learning."
P110,1,KDD,"The paper ""LIDA: A Web Application for Easy and Efficient Dialogue Dataset Creation and Annotation"" is suitable for the KDD conference due to its focus on applying machine learning and natural language processing in dialogue systems. It presents LIDA, a web app simplifying dialogue dataset creation and annotation, integrating ML models as annotation recommenders. The paper also discusses resolving annotator disagreements, crucial for data quality in KDD. Its focus on usability, conflict resolution, and ML integration makes it an excellent fit for KDD."
P111,1,TMLR,"The paper ""Scaling Bayesian Optimization with Deep Learning Surrogates"" combines Bayesian Optimization (BO) with deep learning techniques, specifically Bayesian Neural Networks (BNNs), for scalable and flexible global optimization. BNNs improve BO performance by incorporating domain knowledge and data augmentation techniques, outperforming Gaussian Processes (GPs) on real-world, high-dimensional scientific datasets. The paper's focus on physics-informed priors, scalability, and real-world applicability aligns well with the TMLR conference's objectives."
P112,0,na,na
P113,0,na,na
P114,0,na,na
P115,0,na,na
P116,1,CVPR,"The research paper introduces Top-k, an improved decision tree algorithm that boosts accuracy and scalability in high-dimensional datasets, overcoming limitations of traditional methods like ID3, C4.5, and CART. Top-k offers a flexible control mechanism for exploration-exploitation trade-offs by considering the top k features as potential split candidates. The authors provide theoretical analysis and empirical evidence supporting Top-k's superiority, making it suitable for the Computer Vision and Pattern Recognition (CVPR) conference, which emphasizes efficient algorithms and interpretable models."
P117,1,EMNLP,"The paper investigates the linear representation of word relevance for image tagging. It demonstrates a main direction in word vector space, where relevant tags rank higher than irrelevant ones. The Fast0Tag model is introduced for efficient image tagging. The paper's novel approach to image tagging using word vectors is validated through neural network experiments, making it relevant for EMNLP, a conference focusing on NLP and computational linguistics."
P118,0,na,na
P119,0,na,na
P120,0,na,na
P121,1,NEURIPS,"The paper proposes a novel intervention, ""plasticity injection,"" to address plasticity loss in deep RL, offering an efficient and adaptable solution. It introduces a diagnostic framework, dynamic expansion mechanism, and demonstrates significant improvements in performance and computational efficiency. Keywords align with NeurIPS themes, and the paper's rigorous analysis supports its relevance."
P122,1,KDD,"The research paper proposes a machine learning-based precipitation nowcasting system using the U-Net architecture. It compares the system's predictions with ground-based weather station data, utilizing satellite data and addressing radar coverage gaps in Siberia and the Ural federal districts of Russia. The paper aims to enhance weather forecasting services, aligning with KDD's focus on data mining and knowledge discovery."
P123,1,EMNLP,"The proposed research, focusing on sentiment analysis and emotion detection via a novel emoji-based pre-training method, significantly enhances NLP model performance. The DeepMoji model, pre-trained using a diverse set of emojis, demonstrates superior results and can be generalized across five domains. This aligns well with the EMNLP conference's focus on empirical NLP methods, practical applications, and resource sharing. Keywords include text, layers, word, benchmark, labels, sentiment, approach, tasks."
P124,0,na,na
P125,1,EMNLP,"The ""Conditional Adversarial Filtering for Commonsense Reasoning"" paper is suitable for EMNLP due to its innovative approach to commonsense reasoning, use of advanced machine learning techniques, and significant improvement over existing benchmarks like SWAG and HELLASWAG. The paper presents a new conditional adversarial filtering mechanism, improves sentence plausibility by considering discourse connectives, and discusses pre-training and fine-tuning of a sentence generator model. Comparisons with existing models offer insights into commonsense reasoning model performance."
P126,1,NEURIPS,"The research paper proposes a novel causal inference algorithm for complex real-world scenarios with latent confounders, using a single proxy variable and cross-moments. It demonstrates superior performance in simulations and real-world applications, outperforming existing methods. The paper's focus on robustness and high-dimensionality aligns with NEURIPS' interdisciplinary focus, and its potential applications span healthcare, economics, and social sciences. The paper's theoretical foundations and empirical validation support its reliability and potential impact."
P127,1,KDD,"The research paper, focusing on the interplay of machine learning and privacy, is a suitable submission for the KDD conference. It presents an innovative approach to countering privacy-invasive machine learning systems, advocating for participatory design of anti-surveillance technologies. The paper aligns with KDD's focus on responsible and ethical data practices, offering practical examples of obfuscation and data withholding methods. It also discusses the role of machine learning scientists in addressing these issues and emphasizes societal impact and regulation, aligning with KDD's commitment to fostering responsible data mining."
P128,1,EMNLP,"""End-to-End Discourse Deixis Resolution in Dialogue"" is an appropriate submission for the EMNLP conference due to its focus on natural language processing. The paper investigates discourse deixis resolution, a lesser-explored NLP task, and discusses its contrast with entity coreference resolution. The authors propose enhancements to a span-based model for entity coreference to boost discourse deixis resolution performance in dialogue. The paper also presents an initial analysis of the model, claimed to be the first for a state-of-the-art span-based discourse deixis resolver, aligning with the conference's objectives."
P129,0,na,na
P130,0,na,na
P131,1,CVPR,"The proposed research paper, focusing on unsupervised learning and 3D human motion prediction, utilizes ImageNet-pretrained models, VAEs, and CNNs to address weaknesses in feature extraction and aggregation. It addresses this through a two-step approach, improving disentanglement scores, and employing simulation datasets and classification layers. This innovative contribution to computer vision aligns well with the research objectives and scope of the CVPR conference, making it an ideal fit."
P132,0,na,na
P133,0,na,na
P134,0,na,na
P135,1,NEURIPS,"The research paper proposes a novel extragradient approach for solving decentralized stochastic Minty variational inequalities and saddle-point problems in GANs. It presents this method for flexible network topologies and communication constraints, demonstrating its convergence rate and effectiveness through numerical experiments. The paper's contributions, focusing on distributed learning, are relevant to the NeurIPS community."