Synthetic Data Generation for Minimum-Exposure Navigation in a Time-Varying Environment using Generative AI Models

Drug discovery is a multi-stage process, often beginning with the identification of active molecules from a high-throughput screen or machine learning model. Once structure activity relationship trends become well established, identifying new analogs with better properties is important. Synthesizing these new compounds is a logical next step, and is key to research groups that have a synthetic chemistry team or external collaborators. Generative machine learning models have become widely adopted to generate new molecules and explore molecular space, with the goal of discovering novel compounds with desires properties. These generative models have been composed from recurrent neural networks (RNNs), Variational Autoencoders (VAEs), and Generative Adversarial Networks (GANs) and are often combined with transfer learning or scoring of physicochemical properties to steer generative design. While these generative models have proven useful in generating new molecular libraries, often they are not capable of addressing a wide variety of potential problems, and often converge into similar molecular space when combined with a scoring function for desired properties. In addition, generated compounds are often not synthetically feasible, reducing their capabilities outside of virtual composition and limiting their usefulness in real-world scenarios. Here we introduce a suite of automated tools called MegaSyn representing 3 components: a new hill-climb algorithm which makes use of SMILES-based RNN generative models, analog generation software, and retrosynthetic analysis coupled with fragment analysis to score molecules for their synthetic feasibility. We now describe the development and testing of this suite of tools and propose how they might be used to optimize molecules or prioritize promising lead compounds using test case examples.

Mechanical metamaterials - engineered materials whose behavior is defined by their internal structure rather than their base material - have received great at- tention for the unusual properties they can exhibit. However, while the forward problem of predicting the mechanical behavior of a given metamaterial can be reliably done through simulation or experiments, the inverse problem - designing metamaterials with specific properties - remains difficult, especially for heteroge- neous metamaterials composed of multiple unit cells. This dissertation explores the use of generative machine learning to address this inverse design problem. The first part of the work investigates how inverse and generative models can be conditioned to produce unit cells with prescribed mechanical properties and how they can be used to generate multiple valid solutions for the same target. This is examined using a curved-beam metamaterial, which allows a compact geometric representation without complex design restrictions. The effect these restrictions have on the generative models is investigated in the second part. For this, a number of generative models—Variational Autoencoders, Generative Adversarial Networks, and Diffusion Models—are evaluated on a kirigami metamaterial where dependencies between the cuts cause intricate design constraints. The final part covers the challenges that arise when generative models are applied to heteroge- neous metamaterials, in which multiple unit cells are combined to create spatially varying mechanical properties. The results show that the applicability of generative models for the design of metamaterials strongly depends on the choice of geometric parameterization. Generative models make assumptions about their input data that generally hold for images, but can be invalid for metamaterials when complex dependencies be- tween the input parameters exist. B-splines can provide a representation without those dependencies and are therefore well suited to be paired up with generative models. Furthermore, it is also important how these geometrical parameters are sampled to create the training data for the generative models, as this sampling also affects the distribution of the properties in the training data. Overall, while generative machine learning models prove effective for the inverse design of mechanical metamaterials their successful application requires careful parameterization and selection of the training data.

BackgroundInsufficient patient accrual is a major challenge in clinical trials and can result in underpowered studies, as well as exposing study participants to toxicity and additional costs, with limited scientific benefit. Real-world data can provide external controls, but insufficient accrual affects all arms of a study, not just controls. Studies that used generative models to simulate more patients were limited in the accrual scenarios considered, replicability criteria, number of generative models, and number of clinical trials evaluated.ObjectiveThis study aimed to perform a comprehensive evaluation on the extent generative models can be used to simulate additional patients to compensate for insufficient accrual in clinical trials.MethodsWe performed a retrospective analysis using 10 datasets from 9 fully accrued, completed, and published cancer trials. For each trial, we removed the latest recruited patients (from 10% to 50%), trained a generative model on the remaining patients, and simulated additional patients to replace the removed ones using the generative model to augment the available data. We then replicated the published analysis on this augmented dataset to determine if the findings remained the same. Four different generative models were evaluated: sequential synthesis with decision trees, Bayesian network, generative adversarial network, and a variational autoencoder. These generative models were compared to sampling with replacement (ie, bootstrap) as a simple alternative. Replication of the published analyses used 4 metrics: decision agreement, estimate agreement, standardized difference, and CI overlap.ResultsSequential synthesis performed well on the 4 replication metrics for the removal of up to 40% of the last recruited patients (decision agreement: 88% to 100% across datasets, estimate agreement: 100%, cannot reject standardized difference null hypothesis: 100%, and CI overlap: 0.8-0.92). Sampling with replacement was the next most effective approach, with decision agreement varying from 78% to 89% across all datasets. There was no evidence of a monotonic relationship in the estimated effect size with recruitment order across these studies. This suggests that patients recruited earlier in a trial were not systematically different than those recruited later, at least partially explaining why generative models trained on early data can effectively simulate patients recruited later in a trial. The fidelity of the generated data relative to the training data on the Hellinger distance was high in all cases.ConclusionsFor an oncology study with insufficient accrual with as few as 60% of target recruitment, sequential synthesis can enable the simulation of the full dataset had the study continued accruing patients and can be an alternative to drawing conclusions from an underpowered study. These results provide evidence demonstrating the potential for generative models to rescue poorly accruing clinical trials, but additional studies are needed to confirm these findings and to generalize them for other diseases.

The availability of fine grained time series data is a pre-requisite for research in smart-grids. While data for transmission systems is relatively easily obtainable, issues related to data collection, security and privacy hinder the widespread public availability/accessibility of such datasets at the distribution system level. This has prevented the larger research community from effectively applying sophisticated machine learning algorithms to significantly improve the distribution-level accuracy of predictions and increase the efficiency of grid operations. Synthetic dataset generation has proven to be a promising solution for addressing data availability issues in various domains such as computer vision, natural language processing and medicine. However, its exploration in the smart grid context remains unsatisfactory. Previous works have tried to generate synthetic datasets by modeling the underlying system dynamics: an approach which is difficult, time consuming, error prone and often times infeasible in many problems. In this work, we propose a novel data-driven approach to synthetic dataset generation by utilizing deep generative adversarial networks (GAN) to learn the conditional probability distribution of essential features in the real dataset and generate samples based on the learned distribution. To evaluate our synthetically generated dataset, we measure the maximum mean discrepancy (MMD) between real and synthetic datasets as probability distributions, and show that their sampling distance converges. To further validate our synthetic dataset, we perform common smart grid tasks such as k-means clustering and short-term prediction on both datasets. Experimental results show the efficacy of our synthetic dataset approach: the real and synthetic datasets are indistinguishable by solely examining the output of these tasks.

With the ongoing trend towards digitisation, vast amounts of, often very fine-grained, data are being collected. The ultimate goal is to capture and understand the behaviour of a system, such as the traffic in a city. However, making sense of such data is not straightforward due to its high level of detail and complex dependencies in time and space. Exploring heuristic approaches is essential to arrive at data representations that enable better insights into the underlying system dynamics by zooming out from the detail. In this paper, a novel approach for representing and reasoning about traffic state transition behaviour via a multitude of parameterised Markov chains models, cleverly designed to fit in a cascade, is proposed. The benefits of working with a multitude of individual Markov chains are outlined and subsequently, it is illustrated how to combine them into daily transition graphs such that their graph representation can be exploited to extract insights about daily traffic behaviour. In addition, targeting context-specific studies, an alternative approach is introduced combining in a dynamic fashion a cascade of Markov chains covering longer and overlapping time windows. A recursive algorithm is conceived and validated allowing to exploit this cascade structure for computing state transition probabilities over time. The potential of the proposed approach for mining traffic state transitions is demonstrated on a use case derived from real-world data.

Forecasting Parapapillary atrophy (PPA), i.e., a symptom related to most irreversible eye diseases, provides an alarm for implementing an intervention to slow down the disease progression at early stage. A key question for this forecast is: how to fully utilize the historical data (e.g., retinal image) up to the current stage for future disease prediction? In this paper, we provide an answer with a novel framework, namely Disease Forecast via Progression Learning (DFPL), which exploits the irreversibility prior (i.e., cannot be reversed once diagnosed). Specifically, based on this prior, we decompose two factors that contribute to the prediction of the future disease: i) the current disease label given the data (retinal image, clinical attributes) at present and ii) the future disease label given the progression of the retinal images that from the current to the future. To model these two factors, we introduce the current and progression predictors in DFPL, respectively. In order to account for the degree of progression of the disease, we propose a temporal generative model to accurately generate the future image and compare it with the current one to get a residual image. The generative model is implemented by a recurrent neural network, in order to exploit the dependency of the historical data. To verify our approach, we apply it to a PPA in-house dataset and it yields a significant improvement (e.g., 4.48% of accuracy; 3.45% of AUC) over others. Besides, our generative model can accurately localize the disease-related regions.

This thesis deals with recurrent neural networks, a particular class of artificial neural networks which can learn a generative model of input sequences. The input is mapped, through a feedback loop and a non-linear activation function, into a hidden state, which is then projected into the output space, obtaining either a probability distribution or the new input for the next time-step. This work consists mainly of two parts: a theoretical study for helping the understanding of recurrent neural networks framework, which is not yet deeply investigated, and their application to non-linear prediction problems, since recurrent neural networks are really powerful models suitable for solving several practical tasks in different fields. For what concerns the theoretical part, we analyse the weaknesses of state-of-the-art models and tackle them in order to improve the performance of a recurrent neural network. Firstly, we contribute in the understanding of the dynamical properties of a recurrent neural network, highlighting the close relation between the definition of stable limit cycles and the echo state property of an echo state network. We provide sufficient conditions for the convergence of the hidden state to a trajectory, which is uniquely determined by the input signal, independently of the initial states. This may help extend the memory of the network and increase the design options for the network. Moreover, we develop a novel approach to address the main problem in training recurrent neural networks, the so-called vanishing gradient problem. Our new method allows us to train a very simple recurrent neural network, making the gradient not to vanish even after many time-steps. Exploiting the singular value decomposition of the vanishing factors in the gradient and random matrices theory, we find that the singular values have to be confined in a narrow interval and derive conditions about their root mean square value. Then, we also improve the efficiency of the training of a recurrent neural network, defining a new method for speeding up this process. Thanks to a least square regularization, we can initialize the parameters of the network, in order to set them closer to the minimum and running fewer epochs of classical training algorithms. Moreover, it is also possible to completely train the network with our initialization method, running more iterations of it without losing in performance with respect to classical training algorithms. Finally, it is also possible to use it as a real-time learning algorithm, adjusting the parameters to the new data through one iteration of our initialization. In the last part of this thesis, we apply recurrent neural networks to non-linear prediction problems. We consider prediction of numerical sequences, estimating the following input choosing it from a probability distribution. We study an automatic text generation problem, where we need to predict the following character in order to compose words and sentences, and a path prediction of walking mobile users in the central area of a city, as a sequence of crossroads. Then, we analyse the prediction of video frames, discovering a wide range of applications related to the prediction of movements. We study the collision problem of bouncing balls, taking into account only the sequence of video frames without any knowledge about the physical characteristics of the problem, and the distribution over days of mobile user in a city and in a whole region. Finally, we address the state-of-the-art problem of missing data imputation, analysing the incomplete spectrogram of audio signals. We restore audio signals with missing time-frequency data, demonstrating via numerical experiments that a performance improvement can be achieved involving recurrent neural networks.

Deep generative models (GMs) have transformed the exploration of drug-like chemical space (CS) by generating novel molecules through complex, nontransparent processes, bypassing direct structural similarity. This review examines five key architectures for CS exploration: recurrent neural networks (RNNs), variational autoencoders (VAEs), generative adversarial networks (GANs), normalizing flows (NF), and Transformers. It discusses molecular representation choices, training strategies for focused CS exploration, evaluation criteria for CS coverage, and related challenges. Future directions include refining models, exploring new notations, improving benchmarks, and enhancing interpretability to better understand biologically relevant molecular properties.

The informed sampler: A discriminative approach to Bayesian inference in generative computer vision models

In the past few years, de novo molecular design has increasingly been using generative models from the emergent field of Deep Learning, proposing novel compounds that are likely to possess desired properties or activities. De novo molecular design finds applications in different fields ranging from drug discovery and materials sciences to biotechnology. A panoply of deep generative models, including architectures as Recurrent Neural Networks, Autoencoders, and Generative Adversarial Networks, can be trained on existing data sets and provide for the generation of novel compounds. Typically, the new compounds follow the same underlying statistical distributions of properties exhibited on the training data set Additionally, different optimization strategies, including transfer learning, Bayesian optimization, reinforcement learning, and conditional generation, can direct the generation process toward desired aims, regarding their biological activities, synthesis processes or chemical features. Given the recent emergence of these technologies and their relevance, this work presents a systematic and critical review on deep generative models and related optimization methods for targeted compound design, and their applications.

Sequence classification is critical in the data mining communities. It becomes more challenging when the class distribution is imbalanced, which occurs in many real-world applications. Oversampling algorithms try to re-balance the skewed class by generating synthetic data for minority classes, but most of existing oversampling approaches could not consider the temporal structure of sequences, or handle multivariate and long sequences. To address these problems, this paper proposes a novel oversampling algorithm based on the 'generative' models of sequences. In particular, a recurrent neural network was employed to learn the generative mechanics for sequences as representations for the corresponding sequences. These generative models are then utilized to form a kernel to capture the similarity between different sequences. Finally, oversampling is performed in the kernel feature space to generate synthetic data. The proposed approach can handle highly imbalanced sequential data and is robust to noise. The competitiveness of the proposed approach is demonstrated by experiments on both synthetic data and benchmark data, including univariate and multivariate sequences.

The de novo drug design based on SMILES format is a typical sequence-processing problem. Previous methods based on recurrent neural network (RNN) exhibit limitation in capturing long-range dependency, resulting in a high invalid percentage in generated molecules. Recent studies have shown the potential of Transformer architecture to increase the capacity of handling sequence data. In this work, the encoder module in the Transformer is used to build a generative model. First, we train a Transformer-encoder-based generative model to learn the grammatical rules of known drug molecules and a predictive model to predict the activity of the molecules. Subsequently, transfer learning and reinforcement learning were used to fine-tune and optimize the generative model, respectively, to design new molecules with desirable activity. Compared with previous RNN-based methods, our method has improved the percentage of generating chemically valid molecules (from 95.6 to 98.2%), the structural diversity of the generated molecules, and the feasibility of molecular synthesis. The pipeline is validated by designing inhibitors against the human BRAF protein. Molecular docking and binding mode analysis showed that our method can generate small molecules with higher activity than those carrying ligands in the crystal structure and have similar interaction sites with these ligands, which can provide new ideas and suggestions for pharmaceutical chemists.

We propose a dynamic latent variable model for learning latent bases from time varying, non-negative data. We take a probabilistic approach to modeling the temporal dependence in data by introducing a dynamic Dirichlet prior—a Dirichlet distribution with dynamic parameters. This new distribution allows us to assure non-negativity and avoid intractability when sequential updates are performed otherwise encountered in using Dirichlet prior. We refer to the proposed model as the Dirichlet latent variable model DLVM. We develop an expectation maximization algorithm for the proposed model, and also derive a maximum a posteriori estimate of the parameters. Furthermore, we connect the proposed DLVM to two popular latent basis learning methods—probabilistic latent component analysis PLCA and non-negative matrix factorization NMF. We show that 1 PLCA is a special case of our DLVM, and 2 DLVM can be interpreted as a dynamic version of NMF. The usefulness of DLVM is demonstrated for three audio processing applications—speaker source separation, denoising, and bandwidth expansion. To this end, a new algorithm for source separation is also proposed. Through extensive experiments on benchmark databases, we show that the proposed model outperforms several relevant existing methods in all three applications.

Crop Recommendation Systems powered by machine learning have transformed agriculture by offering farmers data-driven insights to make informed decisions about optimal crop selection. In the existing system, the Support Vector Machine (SVM) algorithm has been employed due to its high prediction accuracy for recommending suitable crops based on various factors such as soil quality, weather conditions, and historical crop data. However, by incorporating the Long Short-Term Memory (LSTM) algorithm, the system can further enhance the accuracy and efficiency of predictions. LSTM’s ability to capture temporal dependencies in data allows for more reliable predictions over time, accounting for trends and patterns that improve crop recommendations. Additionally, the integration of Ant Colony Optimization (ACO) aids in identifying the most profitable regions for farming by analyzing the dataset to predict areas with the highest potential yield. This combination of advanced machine learning techniques significantly improves the overall effectiveness of crop recommendation systems, enabling farmers to maximize productivity and profitability.

Phishing attacks continue to be a significant threat to internet users, necessitating the development of advanced detection systems. This study explores the efficacy of a Bidirectional Gated Recurrent Unit (BiGRU) model combined with feature selection techniques for detecting phishing websites. The dataset used for this research is sourced from the UCI Machine Learning Repository, specifically the Phishing Websites dataset. This approach involves cleaning and preprocessing the data, then normalizing features and employing feature selection to identify the most relevant attributes for classification. The BiGRU model, known for its ability to capture temporal dependencies in data, is then applied. To ensure robust evaluation, we utilized cross-validation, dividing the data into five folds. The experimental results are highly promising, demonstrating a Mean Accuracy, Mean Precision, Mean Recall, Mean F1 Score, and Mean AUC of 1.0. These results indicate the model's exceptional performance distinguishing between phishing and legitimate websites. This study highlights the potential of combining BiGRU models with feature selection and cross-validation to create highly accurate phishing detection systems, providing a reliable solution to enhance cybersecurity measures.