Is there a foreign language describing protein sequences and structures simultaneously? Protein structures, represented by continuous 3D points, have long posed a challenge due to the contrasting modeling paradigms of discrete sequences. We introduce FoldTokenizer to represent protein sequence-structure as discrete symbols. This approach involves projecting residue types and structures into a discrete space, guided by a reconstruction loss for information preservation. We name the learned discrete symbols as FoldToken, and the sequence of FoldTokens serves as a new protein language, transforming the protein sequence-structure into a unified modality. We apply the created protein language on general backbone inpainting task, building the first GPT-style model (FoldGPT) for sequence-structure co-generation with promising results. Key to our success is the substantial enhancement of the vector quantization module, Soft Conditional Vector Quantization (SoftCVQ).

Spatio-temporal predictive learning plays a crucial role in self-supervised learning, with wide-ranging applications across a diverse range of fields. Previous approaches for temporal modeling fall into two categories: recurrent-based and recurrent-free methods. The former, while meticulously processing frames one by one, neglect short-term spatio-temporal information redundancies, leading to inefficiencies. The latter naively stack frames sequentially, overlooking the inherent temporal dependencies. In this paper, we re-examine the two dominant temporal modeling approaches within the realm of spatio-temporal predictive learning, offering a unified perspective. Building upon this analysis, we introduce USTEP (Unified Spatio-TEmporal Predictive learning), an innovative framework that reconciles the recurrent-based and recurrent-free methods by integrating both micro-temporal and macro-temporal scales. Extensive experiments on a wide range of spatio-temporal predictive learning demonstrate that USTEP achieves significant improvements over existing temporal modeling approaches, thereby establishing it as a robust solution for a wide range of spatio-temporal applications.

Molecule inverse folding has been a long-standing challenge in chemistry and biology, with the potential to revolutionize drug discovery and material science. Despite specified models have been proposed for different small- or macro-molecules, few have attempted to unify the learning process, resulting in redundant efforts. Complementary to recent advancements in molecular structure prediction, such as RoseTTAFold All-Atom and AlphaFold3, we propose the unified model UniIF for the inverse folding of all molecules. We do such unification in two levels: 1) Data-Level: We propose a unified block graph data form for all molecules, including the local frame building and geometric feature initialization. 2) Model-Level: We introduce a geometric block attention network, comprising a geometric interaction, interactive attention and virtual long-term dependency modules, to capture the 3D interactions of all molecules. Through comprehensive evaluations across various tasks such as protein design, RNA design, and material design, we demonstrate that our proposed method surpasses state-of-the-art methods on all tasks. UniIF offers a versatile and effective solution for general molecule inverse folding.

Recent research has been remarkable in recurrent neural networks (RNNs) on sequence-to-sequence problems for image caption, and promising in convolutional neural networks (CNNs) on spatial analysis problems for image detection and sematic segmentation problems. In this paper, based on recurrent neural networks and convolutional neural networks, we propose a fusion neural network architecture named F-Net to deal with vehicle trajectory prediction on highway and urban scenarios in autonomous driving applications. The novelty of the proposed method is the attention mechanism that affects effectively in the progress of both RNN and CNN feature extraction. Besides, our sufficient usage of raw sensor data protects scene texture information of environment and interaction among surrounding vehicles. Experimental results on the nuScene dataset show that our proposed method outperforms the state-of-the-art methods.

A large number of color filter arrays (CFAs), periodic or aperiodic, have been proposed. To reconstruct images from all different CFAs and compare their imaging quality, a universal demosaicking method is needed. This paper proposes a new universal demosaicking method based on inter-pixel chrominance capture and optimal demosaicking transformation. It skips the commonly used step to estimate the luminance component at each pixel, and thus, avoids the associated estimation error. Instead, we directly use the acquired CFA color intensity at each pixel as an input component. Two independent chrominance components are estimated at each pixel based on the inter-pixel chrominance in the window, which is captured with the difference of CFA color values between the pixel of interest and its neighbors. Two mechanisms are employed for the accurate estimation: distance-related and edge-sensing weighting to reflect the confidence levels of the inter-pixel chrominance components, and pseudoinverse-based estimation from the components in a window. Then from the acquired CFA color component and two estimated chrominance components, the three primary colors are reconstructed by a linear color transform, which is optimized for the least transform error. Our experiments show that the proposed method is much better than other published universal demosaicking methods.

For high light sensitivity, new CFA designs use panchromatic pixels, aka white pixels, that no visible spectrum energy is filtered. Kodak's CFA2.0 has 50% white pixels, but the demosaicking performance is not good. We present in this work a set of new color filter arrays (CFA) of high light sensitivity and high demosaicking performance which were obtained by using a CFA design methodology in the frequency domain. The new patterns are of size 5×5 and come from the same frequency structure, which has one luma in the base band at (0, 0) and four chromas (two conjugate pairs) placed at (4π/5, 2π/5), (- 4π/5, - 2π/5), (2π/5, - 4π/5) and (- 2π/5, 4π/5), respectively. The new patterns are optimized to have only white (panchromatic) and three primary color pixels and the pixels are found to be 40% white, 20% red, 20% green and 20% blue by pixel color constrained optimization. Our demosaicking experiments show that our new CFA patterns outperform Kodak CFA2.0 in both objective and subjective quality.

We develop an algorithm RSFA to perform nonlinear blind source separation with temporal constraints. The algorithm is based on slow feature analysis using random Fourier features for shift invariant kernels, followed by a selection procedure to obtain the sought-after signals. This method not only obtains remarkable results in a short computing time, but also excellently handles situations where there are multiple types of mixtures. In kernel methods, since the problem is unsupervised, the need of multiple kernels is ubiquitous. Experiments on music excerpts illustrate the strong performance of our method.

In multi-instance learning, there are two kinds of prediction failure, i.e., false negative and false positive. Current research mainly focus on avoding the former. We attempt to utilize the geometric distribution of instances inside positive bags to avoid both the former and the latter. Based on kernel principal component analysis, we define a projection constraint for each positive bag to classify its constituent instances far away from the separating hyperplane while place positive instances and negative instances at opposite sides. We apply the Constrained Concave-Convex Procedure to solve the resulted problem. Empirical results demonstrate that our approach offers improved generalization performance.

Peculiarity-oriented mining (POM) is a new data mining method consisting of peculiar data identification and peculiar data analysis. Peculiarity factor (PF) and local peculiarity factor (LPF) are important concepts employed to describe the peculiarity of points in the identification step. One can study the notions at both attribute and record levels. In this paper, a new record LPF called distance based record LPF (D-record LPF) is proposed, which is defined as the sum of distances between a point and its nearest neighbors. It is proved mathematically that D-record LPF can characterize accurately the probability density function of a continuous m-dimensional distribution. This provides a theoretical basis for some existing distance based anomaly detection techniques. More important, it also provides an effective method for describing the class conditional probabilities in the Bayesian classifier. The result enables us to apply peculiarity analysis for classification problems. A novel algorithm called LPF-Bayes classifier and its kernelized implementation are presented, which have some connection to the Bayesian classifier. Experimental results on several benchmark data sets demonstrate that the proposed classifiers are effective.

Recently increasing attention has been focused on directly optimizing ranking measures and inducing sparsity in learning models. However, few attempts have been made to relate them together in approaching the problem of learning to rank. In this paper, we consider the sparse algorithms to directly optimize the Normalized Discounted Cumulative Gain (NDCG) which is a widely-used ranking measure. We begin by establishing a reduction framework under which we reduce ranking, as measured by NDCG, to the importance weighted pairwise classification. Furthermore, we provide a sound theoretical guarantee for this reduction, bounding the realized NDCG regret in terms of a properly weighted pairwise classification regret, which implies that good performance can be robustly transferred from pairwise classification to ranking. Based on the converted pairwise loss function, it is conceivable to take into account sparsity in ranking models and to come up with a gradient possessing certain performance guarantee. For the sake of achieving sparsity, a novel algorithm named RSRank has also been devised, which performs L1 regularization using truncated gradient descent. Finally, experimental results on benchmark collection confirm the significant advantage of RSRank in comparison with several baseline methods.

This paper aims to conduct a study on the listwise approach to learning to rank. The listwise approach learns a ranking function by taking individual lists as instances and minimizing a loss function defined on the predicted list and the ground-truth list. Existing work on the approach mainly focused on the development of new algorithms; methods such as RankCosine and ListNet have been proposed and good performances by them have been observed. Unfortunately, the underlying theory was not sufficiently studied so far. To amend the problem, this paper proposes conducting theoretical analysis of learning to rank algorithms through investigations on the properties of the loss functions, including consistency, soundness, continuity, differentiability, convexity, and efficiency. A sufficient condition on consistency for ranking is given, which seems to be the first such result obtained in related research. The paper then conducts analysis on three loss functions: likelihood loss, cosine loss, and cross entropy loss. The latter two were used in RankCosine and ListNet. The use of the likelihood loss leads to the development of a new listwise method called ListMLE, whose loss function offers better properties, and also leads to better experimental results.

Peculiarity oriented mining (POM), aiming to discover peculiarity rules hidden in a dataset, is a new data mining method. In the past few years, many results and applications on POM have been reported. However, there is still a lack of theoretical analysis. In this paper, we prove that the peculiarity factor (PF), one of the most important concepts in POM, can accurately characterize the peculiarity of data with respect to the probability density function of a normal distribution, but is unsuitable for more general distributions. Thus, we propose the concept of local peculiarity factor (LPF). It is proved that the LPF has the same ability as the PF for a normal distribution and is the so-called µ-sensitive peculiarity description for general distributions. To demonstrate the effectiveness of the LPF, we apply it to outlier detection problems and give a new outlier detection algorithm called LPF-Outlier. Experimental results show that LPF-Outlier is an effective outlier detection algorithm.

Most existing algorithms for ordinal regression usually seek an orientation for which the projected samples are well separated, and seriate intervals on that orientation to represent the ranks. However, these algorithms only make use of one dimension in the sample space, which would definitely lose some useful information in its complementary subspace. As a remedy, we propose an algorithm framework for ordinal regression which consists of two phases: recursively extracting features from the decreasing subspace and learning a ranking rule from the examples represented by the new features. In this framework, every algorithm that projects samples onto a line can be used as a feature extractor and features with decreasing ranking ability are extracted one by one to make best use of the information contained in the training samples. Experiments on synthetic and benchmark datasets verify the usefulness of our framework.

The effectiveness of any machine learning algorithm depends, to a large extent, on the selection of a good subset of features or attributes. Most existing methods use the syntactic or statistical information of the data, relying on a heuristic criterion to select features. In this paper, we investigate an alternative less-studied approach called user-oriented feature selection by exploiting the domain-specific semantic information. Given any two features, a user is able to express which one is more important based on the semantic consideration. Such user requirements are formally described by a preference relation on the set of features. Algorithms are proposed to construct a subset of features that is most consistent with the user requirements. Their properties and computational complexity are analysed. User-oriented feature selection offers a new view for machine learning and its potentials need to be further investigated and explored.

This paper proposes a complete framework of posterior probability support vector machines (PPSVMs) for weighted training samples using modified concepts of risks, linear separability, margin, and optimal hyperplane. Within this framework, a new optimization problem for unbalanced classification problems is formulated and a new concept of support vectors established. Furthermore, a soft PPSVM with an interpretable parameter v is obtained which is similar to the v-SVM developed by Schölkopf et al., and an empirical method for determining the posterior probability is proposed as a new approach to determine v. The main advantage of an PPSVM classifier lies in that fact that it is closer to the Bayes optimal without knowing the distributions. To validate the proposed method, two synthetic classification examples are used to illustrate the logical correctness of PPSVMs and their relationship to regular SVMs and Bayesian methods. Several other classification experiments are conducted to demonstrate that the performance of PPSVMs is better than regular SVMs in some cases. Compared with fuzzy support vector machines (FSVMs), the proposed PPSVM is a natural and an analytical extension of regular SVMs based on the statistical learning theory.

We have proposed to use the method of principal curves to describe and analyze the interaction among freeway traffic-stream variables and their joint behaviors without utilizing conventional assumptions made on the functional forms of interactions, as in previous studies. As a nonparameter modeling approach, the performance of the proposed method depends only on the data used and involves no assumed knowledge regarding the relationship among the traffic-stream variables. First, we discuss the basic algorithm for data analysis using principal curves and the corresponding data filter algorithm for determining principal curves for application in traffic-steam analysis. Second, a case study is used to compare the performance of the proposed method to that of the classical model proposed by Greenshields; results indicate that the proposed model is better than the classical one in both data accuracy and curve shape. Finally, the traffic-stream models generated with principal curves at different locations and lanes are compared with each others and the three-dimensional traffic-stream models developed from principal curves are discussed. Clearly, our results have demonstrated the feasibility and advantages of applying principal curves in freeway traffic-stream modeling and analysis.