A distinctive representation of image patches in form of features is a key component of many computer vision and robotics tasks, such as image matching, image retrieval, and visual localization. State-of-the-art descriptors, from hand-crafted descriptors such as SIFT to learned ones such as HardNet, are usually high dimensional; 128 dimensions or even more. The higher the dimensionality, the larger the memory consumption and computational time for approaches using such descriptors. In this paper, we investigate multi-layer perceptrons (MLPs) to extract low-dimensional but high-quality descriptors. We thoroughly analyze our method in unsupervised, self-supervised, and supervised settings, and evaluate the dimensionality reduction results on four representative descriptors. We consider different applications, including visual localization, patch verification, image matching and retrieval. The experiments show that our lightweight MLPs achieve better dimensionality reduction than PCA. The lower-dimensional descriptors generated by our approach outperform the original higher-dimensional descriptors in downstream tasks, especially for the hand-crafted ones. The code will be available at https://github.com/PRBonn/descriptor-dr.

When dealing with clinical text classification on a small dataset recent studies have confirmed that a well-tuned multilayer perceptron outperforms other generative classifiers, including deep learning ones. To increase the performance of the neural network classifier, feature selection for the learning representation can effectively be used. However, most feature selection methods only estimate the degree of linear dependency between variables and select the best features based on univariate statistical tests. Furthermore, the sparsity of the feature space involved in the learning representation is ignored. Goal: Our aim is therefore to access an alternative approach to tackle the sparsity by compressing the clinical representation feature space, where limited French clinical notes can also be dealt with effectively. Methods: This study proposed an autoencoder learning algorithm to take advantage of sparsity reduction in clinical note representation. The motivation was to determine how to compress sparse, high-dimensional data by reducing the dimension of the clinical note representation feature space. The classification performance of the classifiers was then evaluated in the trained and compressed feature space. Results: The proposed approach provided overall performance gains of up to 3% for each evaluation. Finally, the classifier achieved a 92% accuracy, 91% recall, 91% precision, and 91% f1-score in detecting the patient's condition. Furthermore, the compression working mechanism and the autoencoder prediction process were demonstrated by applying the theoretic information bottleneck framework.

In this paper, we advocate for two stages in a neural network's decision making process. The first is the existing feed-forward inference framework where patterns in given data are sensed and associated with previously learned patterns. The second stage is a slower reflection stage where we ask the network to reflect on its feed-forward decision by considering and evaluating all available choices. Together, we term the two stages as introspective learning. We use gradients of trained neural networks as a measurement of this reflection. A simple three-layered Multi Layer Perceptron is used as the second stage that predicts based on all extracted gradient features. We perceptually visualize the post-hoc explanations from both stages to provide a visual grounding to introspection. For the application of recognition, we show that an introspective network is 4% more robust and 42% less prone to calibration errors when generalizing to noisy data. We also illustrate the value of introspective networks in downstream tasks that require generalizability and calibration including active learning, out-of-distribution detection, and uncertainty estimation. Finally, we ground the proposed machine introspection to human introspection for the application of image quality assessment.

Differential equations play a pivotal role in modern world ranging from science, engineering, ecology, economics and finance where these can be used to model many physical systems and processes. In this paper, we study two mathematical models of a drug assimilation in the human system using Physics Informed Neural Networks (PINNs). In the first model, we consider the case of single dose of drug in the human system and in the second case, we consider the course of this drug taken at regular intervals. We have used the compartment diagram to model these cases. The resulting differential equations are solved using PINN, where we employ a feed forward multilayer perceptron as function approximator and the network parameters are tuned for minimum error. Further, the network is trained by finding the gradient of the error function with respect to the network parameters. We have employed DeepXDE, a python library for PINNs, to solve the simultaneous first order differential equations describing the two models of drug assimilation. The results show high degree of accuracy between the exact solution and the predicted solution as much as the resulting error reaches10^(-11) for the first model and 10^(-8) for the second model. This validates the use of PINN in solving any dynamical system.

Medical vision-and-language pre-training provides a feasible solution to extract effective vision-and-language representations from medical images and texts. However, few studies have been dedicated to this field to facilitate medical vision-and-language understanding. In this paper, we propose a self-supervised learning paradigm with multi-modal masked autoencoders (M$^3$AE), which learn cross-modal domain knowledge by reconstructing missing pixels and tokens from randomly masked images and texts. There are three key designs to make this simple approach work. First, considering the different information densities of vision and language, we adopt different masking ratios for the input image and text, where a considerably larger masking ratio is used for images. Second, we use visual and textual features from different layers to perform the reconstruction to deal with different levels of abstraction in visual and language. Third, we develop different designs for vision and language decoders (i.e., a Transformer for vision and a multi-layer perceptron for language). To perform a comprehensive evaluation and facilitate further research, we construct a medical vision-and-language benchmark including three tasks. Experimental results demonstrate the effectiveness of our approach, where state-of-the-art results are achieved on all downstream tasks. Besides, we conduct further analysis to better verify the effectiveness of different components of our approach and various settings of pre-training. The source code is available at~\url{https://github.com/zhjohnchan/M3AE}.

Today's artificial intelligence (AI) can do some extraordinary things. Its functionality, though, has very little to do with the way in which a human brain works to achieve the same tasks. For AI to overcome its inherent limitations and advance to artificial general intelligence, we must recognize the differences between the brain and its artificial counterparts. With that in mind, this nine-part series examines the capabilities and limitations of biological neurons and how these relate to machine learning (ML). In the first two parts of this series, we examined how a neuron's slowness makes an ML approach to learning implausible in neurons, and how the fundamental algorithm of the perceptron differs from a biological neuron model involving spikes.

Research articles are being shared in increasing numbers on multiple online platforms. Although the scholarly impact of these articles has been widely studied, the online interest determined by how long the research articles are shared online remains unclear. Being cognizant of how long a research article is mentioned online could be valuable information to the researchers. In this paper, we analyzed multiple social media platforms on which users share and/or discuss scholarly articles. We built three clusters for papers, based on the number of yearly online mentions having publication dates ranging from the year 1920 to 2016. Using the online social media metrics for each of these three clusters, we built machine learning models to predict the long-term online interest in research articles. We addressed the prediction task with two different approaches: regression and classification. For the regression approach, the Multi-Layer Perceptron model performed best, and for the classification approach, the tree-based models performed better than other models. We found that old articles are most evident in the contexts of economics and industry (i.e., patents). In contrast, recently published articles are most evident in research platforms (i.e., Mendeley) followed by social media platforms (i.e., Twitter).

There are several varieties of neural networks, each of which serves a particular function. Frank Rosenblatt invented the perceptron in 1958, making it the earliest neural network. This article has mostly focused on feedforward neural networks, often known as multi-layer perceptrons (MLPs). They are made up of three layers: an input layer, a concealed layer or layers, and an output layer. While these neural networks are also known as MLPs, it is crucial to highlight that they are made up of sigmoid neurons rather than perceptrons, as most real-world issues are nonlinear.

Vision multi-layer perceptrons (MLPs) have shown promising performance in computer vision tasks, and become the main competitor of CNNs and vision Transformers. They use token-mixing layers to capture cross-token interactions, as opposed to the multi-head self-attention mechanism used by Transformers. However, the heavily parameterized token-mixing layers naturally lack mechanisms to capture local information and multi-granular non-local relations, thus their discriminative power is restrained. To tackle this issue, we propose a new positional spacial gating unit (PoSGU). It exploits the attention formulations used in the classical relative positional encoding (RPE), to efficiently encode the cross-token relations for token mixing. It can successfully reduce the current quadratic parameter complexity $O(N^2)$ of vision MLPs to $O(N)$ and $O(1)$. We experiment with two RPE mechanisms, and further propose a group-wise extension to improve their expressive power with the accomplishment of multi-granular contexts. These then serve as the key building blocks of a new type of vision MLP, referred to as PosMLP. We evaluate the effectiveness of the proposed approach by conducting thorough experiments, demonstrating an improved or comparable performance with reduced parameter complexity. For instance, for a model trained on ImageNet1K, we achieve a performance improvement from 72.14\% to 74.02\% and a learnable parameter reduction from $19.4M$ to $18.2M$. Code could be found at https://github.com/Zhicaiwww/PosMLP.

While leading Neural Network architectures for NLP are helmed by Transformers (since 2017, with the paper "Attention Is All you Need"), Computer Vision progress has been led by Convolutional Neural Networks (CNN) ever since AlexNet became the first CNN winner of the ImageNet challenge in 2012 -- though its supremacy has been contested by Vision Transformers (ViT) since 2020. From 2012 onward, many variations of CNN have been developed to tackle different Computer Vision problems, such as Instance Segmentation and Objection Detection. Before the age of CNN, the simple approach to Computer Vision is to treat images pixels as individual features to feed into the deep neural networks (Multi-Layer Perceptrons). Fortunately, the ingenious CNN architecture comes to the rescue. In this article, we will explore basic CNN architecture, and then utilize Transfer Learning with CNN-- borrowing state-of-the-art architectures with pre-trained weights fine-tuned on ImageNet -- to achieve cutting-edge results on predicting a Kaggle's Cat versus Dog dataset.