As an important part of Music Information Retrieval (MIR), Symbolic Music Understanding (SMU) has gained substantial attention, as it can assist musicians and amateurs in learning and creating music. Recently, pre-trained language models have been widely adopted in SMU because the symbolic music shares a huge similarity with natural language, and the pre-trained manner also helps make full use of limited music data. However, the issue of bias, such as sexism, ageism, and racism, has been observed in pre-trained language models, which is attributed to the imbalanced distribution of training data. It also has a significant influence on the performance of downstream tasks, which also happens in SMU. To address this challenge, we propose Adversarial-MidiBERT, a symbolic music understanding model based on Bidirectional Encoder Representations from Transformers (BERT). We introduce an unbiased pre-training method based on adversarial learning to minimize the participation of tokens that lead to biases during training. Furthermore, we propose a mask fine-tuning method to narrow the data gap between pre-training and fine-tuning, which can help the model converge faster and perform better. We evaluate our method on four music understanding tasks, and our approach demonstrates excellent performance in all of them. The code for our model is publicly available at https://github.com/RS2002/Adversarial-MidiBERT.

California would become the first state to require venture capital firms to disclose the race and gender of the founders of the companies they fund, under a bill currently awaiting governor Gavin Newsom's signature. The business community strongly opposes the legislation, characterizing it as an example of bureaucratic overreach. But civil rights groups and female entrepreneurs say it could go a long way toward equalizing opportunity in Silicon Valley, where startup capital overwhelmingly flows to white men. According to the business data firm PitchBook, companies founded by all-female teams accounted for just 2% of venture capital funding last year. Those led by Black women and Latinas received even less, 0.85%, according to a report from Project Diane, a research effort focused on female founders.

The AllInOne training paradigm squeezes a wide range of tasks into a unified model in a multi-task learning manner. However, optimization in multi-task learning is more challenge than single-task learning, as the gradient norm from different tasks may vary greatly, making the backbone overly biased towards one specific task. To address this issue, we propose the task-level backbone-oriented gradient clip paradigm, compared with the vanilla gradient clip method, it has two points of emphasis:1) gradient clip is performed independently for each task. 2) backbone gradients generated from each task are rescaled to the same norm scale. Based on the experimental results, we argue that the task-level backbone-oriented gradient clip paradigm can relieve the gradient bias problem to some extent. We also propose a novel multi-branch data augmentation strategy where conflict augmentations are placed in different branches. Our approach has been shown to be effective and finally achieve 1st place in the Leaderboard A and 2nd place in the Leaderboard B of the CVPR2023 Foundation Model Challenge. It's worth noting that instead of evaluating all three tasks(detection, segmentation and fine-grained classification) in Leaderboard A, the segmentation task is not evaluated in Leaderboard B, in which our team has a huge advantage.

In computer vision, the performance of deep neural networks (DNNs) is highly related to the feature extraction ability, i.e., the ability to recognize and focus on key pixel regions in an image. However, in this paper, we quantitatively and statistically illustrate that DNNs have a serious attention bias problem on many samples from some popular datasets: (1) Position bias: DNNs fully focus on label-independent regions; (2) Range bias: The focused regions from DNN are not completely contained in the ideal region. Moreover, we find that the existing self-attention modules can alleviate these biases to a certain extent, but the biases are still non-negligible. To further mitigate them, we propose a lightweight sub-attention strategy (LSAS), which utilizes high-order sub-attention modules to improve the original self-attention modules. The effectiveness of LSAS is demonstrated by extensive experiments on widely-used benchmark datasets and popular attention networks. We release our code to help other researchers to reproduce the results of LSAS~\footnote{https://github.com/Qrange-group/LSAS}.

In a cluttered online world, few can resist the convenience of an automated ranking when deciding what movie to watch on Netflix or which seafood restaurant looks promising in a Google search. But when it comes to finding a job candidate or someone to do a basic household task, there's often a human toll to letting algorithms do the work. Searches on popular recruiting sites might seem like a neutral way to find prospective candidates, but their underlying technology can reinforce biases by excluding underrepresented groups, including women. For instance, research shows that women receive fewer employment reviews on the popular online freelancing site TaskRabbit compared to men with the same experience--and this lack of reviews can lower the rankings of women in talent search algorithms. "Maybe there is a bias from people who have been traditionally hiring men," explains Himabindu Lakkaraju, an assistant professor at Harvard Business School.

Now a days Artificial intelligence systems are critical for companies that wish to extract value from data by automating and optimizing processes or producing actionable insights but a concept named "Bias problem" comes into picture in situations where ML-based data analytics systems show bias against certain groups of people when predicting the output. The bias problem occurs when the training data which is being fed into a Machine learning model is more inclined towards a certain category of group in the input set. Due to this when the machine is trained with this data, the output generated by Machine learning model is more inclined towards the certain feature which leads to incorrect prediction of output. Bias can make way into algorithms in several ways. AI systems learn to make decisions based on training data, which can include biased human decisions or reflect historical inequities.

Unmanned aerial vehicles (UAVs) mobility enables flexible and customized federated learning (FL) at the network edge. However, the underlying uncertainties in the aerial-terrestrial wireless channel may lead to a biased FL model. In particular, the distribution of the global model and the aggregation of the local updates within the FL learning rounds at the UAVs are governed by the reliability of the wireless channel. This creates an undesirable bias towards the training data of ground devices with better channel conditions, and vice versa. This paper characterizes the global bias problem of aerial FL in large-scale UAV networks. To this end, the paper proposes a channel-aware distribution and aggregation scheme to enforce equal contribution from all devices in the FL training as a means to resolve the global bias problem. We demonstrate the convergence of the proposed method by experimenting with the MNIST dataset and show its superiority compared to existing methods. The obtained results enable system parameter tuning to relieve the impact of the aerial channel deficiency on the FL convergence rate.

Despite my enthusiasm, I do not see AI as a panacea which will transform the world only for "the good". The present state of things may be more akin to the "Sorcerer's Apprentice" section of Fantasia. As revolutionary a tool as these visualization applications may be for any number of end uses, they are small potatoes compared to the ways that AI will more broadly change our world -- in weapons, in metacognition (AI assistants), in medicine and biotechnology, in deepfake / disinformation vs detection arms races, and so on. Great and terrible things are likely to result, as Tesla famously said, "You May Live to See Man-Made Horrors Beyond Your Comprehension." Even in the relatively smaller arena of generative imagery, I see potential storm clouds gathering on the horizon.

Dataset bias has attracted increasing attention recently for its detrimental effect on the generalization ability of fine-tuned models. The current mainstream solution is designing an additional shallow model to pre-identify biased instances. However, such two-stage methods scale up the computational complexity of training process and obstruct valid feature information while mitigating bias. To address this issue, we utilize the representation normalization method which aims at disentangling the correlations between features of encoded sentences. We find it also promising in eliminating the bias problem by providing isotropic data distribution. We further propose Kernel-Whitening, a Nystrom kernel approximation method to achieve more thorough debiasing on nonlinear spurious correlations. Our framework is end-to-end with similar time consumption to fine-tuning. Experiments show that Kernel-Whitening significantly improves the performance of BERT on out-of-distribution datasets while maintaining in-distribution accuracy.

Object navigation tasks require agents to locate specific objects in unknown environments based on visual information. Previously, graph convolutions were used to implicitly explore the relationships between objects. However, due to differences in visibility among objects, it is easy to generate biases in object attention. Thus, in this paper, we propose a directed object attention (DOA) graph to guide the agent in explicitly learning the attention relationships between objects, thereby reducing the object attention bias. In particular, we use the DOA graph to perform unbiased adaptive object attention (UAOA) on the object features and unbiased adaptive image attention (UAIA) on the raw images, respectively. To distinguish features in different branches, a concise adaptive branch energy distribution (ABED) method is proposed. We assess our methods on the AI2-Thor dataset. Compared with the state-of-the-art (SOTA) method, our method reports 7.4%, 8.1% and 17.6% increase in success rate (SR), success weighted by path length (SPL) and success weighted by action efficiency (SAE), respectively.