Partial-label learning (PLL) is a weakly supervised learning problem in which each example is associated with multiple candidate labels and only one is the true label. In recent years, many deep PLL algorithms have been developed to improve model performance. However, we find that some early developed algorithms are often underestimated and can outperform many later algorithms with complicated designs. In this paper, we delve into the empirical perspective of PLL and identify several critical but previously overlooked issues. First, model selection for PLL is non-trivial, but has never been systematically studied. Second, the experimental settings are highly inconsistent, making it difficult to evaluate the effectiveness of the algorithms. Third, there is a lack of real-world image datasets that can be compatible with modern network architectures. Based on these findings, we propose PLENCH, the first Partial-Label learning bENCHmark to systematically compare state-of-the-art deep PLL algorithms. We investigate the model selection problem for PLL for the first time, and propose novel model selection criteria with theoretical guarantees. We also create Partial-Label CIFAR-10 (PLCIFAR10), an image dataset of human-annotated partial labels collected from Amazon Mechanical Turk, to provide a testbed for evaluating the performance of PLL algorithms in more realistic scenarios. Researchers can quickly and conveniently perform a comprehensive and fair evaluation and verify the effectiveness of newly developed algorithms based on PLENCH. We hope that PLENCH will facilitate standardized, fair, and practical evaluation of PLL algorithms in the future.

Weaknesses: The ideas of Laplacian coordinates and differential pooling have been explored in existing works on graph neural networks, e.g., in those works on Spectral-based Convolutional GNN (see [1] below). So technically, in Section 3.3, can you provide comparison not only with standard CNN but also the recent graph neural network models and state the novelty of this work. The idea of adopting AMG is novel and the only work that I am aware of is [29] "Learning Algebraic Multigrid Using Graph Neural Networks," which is recently published in ICML 2020 (please update the reference). However, seeing Section 3.3, the contribution of adopting AMG is not very strong. This work can be stronger, if it explores AMG in greater depth.

The authors theoretically prove that "more skip connections the faster convergence" and "shallow cells benefit faster convergence rate than deep cells". Is there any experimental evidence to verify these claims? However, does pooling operations also have a slower convergence rate than skip connections? In lines 175-178, the authors mentioned that skip connection in shared path and convolution in private path can benefit the Gram matrix singularity of networks. Thus, the convergence rate can be greatly improved.

The International Conference on Machine Learning (ICML) received nearly 5,000 submissions for its 2020 conference and accepted 1,088 papers. Machine Learning Center at Georgia Tech (ML@GT) researchers authored nine accepted papers. The papers explore topics like privacy, semantics in predictive agents, data science, and artificial intelligence. One paper, Boosting Frank-Wolfe by Chasing Gradients, proposes a new state-of-the-art algorithm for constrained optimization, an area already addressed in work accepted in 2019. "I think we're going to see a lot more work moving in the direction of general artificial intelligence, especially work that is trying to combine learning and reasoning," said Le Song, an associate director at ML@GT.

The International Conference on Machine Learning (ICML) is a flagship machine learning conference that in 2020 received 4,990 submissions and managed a pool of 3,931 reviewers and area chairs. Given that the stakes in the review process are high -- the careers of researchers are often significantly affected by the publications in top venues -- we decided to scrutinize several components of the peer-review process in a series of experiments. Specifically, in conjunction with the ICML 2020 conference, we performed three experiments that target: resubmission policies, management of reviewer discussions, and reviewer recruiting. In this post, we summarize the results of these studies. Several leading ML and AI conferences have recently started requiring authors to declare previous submission history of their papers.

With this year's International Conference on Machine Learning (ICML) being over, it is time to have another instalment of this series. Similar to last year's post, I shall cover several papers that caught my attention because of their use of topological concepts--however, unlike last year, I shall not restrict the selection to papers using topological data analysis (TDA). Caveat lector: I might have missed some promising papers. Any suggestions for additions are more than welcome! Please reach out to me via Twitter or e-mail.

This year's virtual ICML conference hosted 10800 attendees from 75 countries. Apparently, the virtual format makes big research conferences such as ICML more accessible to the AI community all over the world. With almost 5000 research papers submitted to ICML 2020 and an acceptance rate of 21.8%, a total of 1088 papers were presented at the conference. As usual, the Outstanding Papers awards were given to exemplary papers at this year's ICML. To help you stay aware of the most prominent AI research breakthroughs, we've summarized the key ideas of these papers.

\textbf{G}raph \textbf{C}onvolutional \textbf{N}etwork (\textbf{GCN}) is widely used in graph data learning tasks such as recommendation. However, when facing a large graph, the graph convolution is very computationally expensive thus is simplified in all existing GCNs, yet is seriously impaired due to the oversimplification. To address this gap, we leverage the \textit{original graph convolution} in GCN and propose a \textbf{L}ow-pass \textbf{C}ollaborative \textbf{F}ilter (\textbf{LCF}) to make it applicable to the large graph. LCF is designed to remove the noise caused by exposure and quantization in the observed data, and it also reduces the complexity of graph convolution in an unscathed way. Experiments show that LCF improves the effectiveness and efficiency of graph convolution and our GCN outperforms existing GCNs significantly. Codes are available on \url{https://github.com/Wenhui-Yu/LCFN}.

Stanford University had a notable presence at the 37th International Conference on Machine Learning (ICML) last week, with a leading number of citations and papers, ahead of other machine learning powerhouses MIT, U.C. Berkeley, Carnegie Mellon and Princeton. The ICML 2020, which was held from July 12 to July 18, was originally scheduled to take place in Vienna, Austria, but due to the COVID-19 pandemic, the conference was moved to a virtual format. The ICML is a leading academic conference on machine learning, at which academic and industry teams from around the world present cutting-edge research in the field. This year, only 1,088 papers were selected from 4,990 submissions, an all-time low acceptance of 21.8%, according to an article published on Medium by Gleb Chuvpilo, a partner of the firm Thundermark Capital, which invests in AI and robotics. Stanford also proved itself to be a leader in Artificial Intelligence research relative to industry representatives, second only to Google, and ahead of leading companies like Microsoft, Facebook and IBM.

Organizers of the 37th International Conference on Machine Learning (ICML) have announced their Outstanding Paper awards, recognizing papers from the current conference that are "strong representatives of solid theoretical and empirical work in our field." A total of 1,088 papers out of 4,990 submissions made it to the prestigious machine learning conference. The acceptance rate of 21.8 percent is slightly lower than 2019's 22.6 percent (774 accepted papers from 3,424 submissions), and it seems likely the drastic increase in submissions helped contribute to this. Authors: Haggai Maron, Or Litany, Gal Chechik, Ethan Fetaya Institutions: NVIDIA Research, Stanford University, Bar Ilan University Abstract: Learning from unordered sets is a fundamental learning setup, recently attracting increasing attention. Research in this area has focused on the case where elements of the set are represented by feature vectors, and far less emphasis has been given to the common case where set elements themselves adhere to their own symmetries.