Time series anomaly detection (TSAD) is an evolving area of research motivated by its critical applications, such as detecting seismic activity, sensor failures in industrial plants, predicting crashes in the stock market, and so on. Across domains, anomalies occur significantly less frequently than normal data, making the F1-score the most commonly adopted metric for anomaly detection. However, in the case of time series, it is not straightforward to use standard F1-score because of the dissociation between ‘time points’ and ‘time events’. To accommodate this, anomaly predictions are adjusted, called as point adjustment (PA), before the F1-score evaluation. However, these adjustments are heuristics-based, and biased towards true positive detection, resulting in over-estimated detector performance. In this work, we propose an alternative adjustment protocol called "Balanced point adjustment" (BA). It addresses the limitations of existing point adjustment methods and provides guarantees of fairness backed by axiomatic definitions of TSAD evaluation. Code and implementation details: https://github.com/summukhe/balanced_f1score.

Large pre-trained models excel in zero/few-shot learning for language and vision tasks but face challenges in multivariate time series (TS) forecasting due to diverse data characteristics. Consequently, recent research efforts have focused on developing pre-trained TS forecasting models. These models, whether built from scratch or adapted from large language models (LLMs), excel in zero/few-shot forecasting tasks. However, they are limited by slow performance, high computational demands, and neglect of cross-channel and exogenous correlations. To address this, we introduce Tiny Time Mixers (TTM), a compact model (starting from 1M parameters) with effective transfer learning capabilities, trained exclusively on public TS datasets. TTM, based on the light-weight TSMixer architecture, incorporates innovations like adaptive patching, diverse resolution sampling, and resolution prefix tuning to handle pre-training on varied dataset resolutions with minimal model capacity. Additionally, it employs multi-level modeling to capture channel correlations and infuse exogenous signals during fine-tuning. TTM outperforms existing popular benchmarks in zero/few-shot forecasting by (4-40\%), while reducing computational requirements significantly. Moreover, TTMs are lightweight and can be executed even on CPU-only machines, enhancing usability and fostering wider adoption in resource-constrained environments. The model weights for reproducibility and research use are available at https://huggingface.co/ibm/ttm-research-r2/, while enterprise-use weights under the Apache license can be accessed as follows: the initial TTM-Q variant at https://huggingface.co/ibm-granite/granite-timeseries-ttm-r1, and the latest variants (TTM-B, TTM-E, TTM-A) weights are available at https://huggingface.co/ibm-granite/granite-timeseries-ttm-r2. The source code for the TTM model along with the usage scripts are available at https://github.com/ibm-granite/granite-tsfm/tree/main/tsfm_public/models/tinytimemixer

With the growing adoption of AI, trust and explainability have become critical which has attracted a lot of research attention over the past decade and has led to the development of many popular AI explainability libraries such as AIX360, Alibi, OmniXAI, etc. Despite that, applying explainability techniques in practice often poses challenges such as lack of consistency between explainers, semantically incorrect explanations, or scalability. Furthermore, one of the key modalities that has been less explored, both from the algorithmic and practice point of view, is time-series. Several application domains involve time-series including Industry 4.0, asset monitoring, supply chain or finance to name a few.

Selecting the right set of hyperparameters is crucial in time series forecasting. The classical temporal cross-validation framework for hyperparameter optimization (HPO) often leads to poor test performance because of a possible mismatch between validation and test periods. To address this test-validation mismatch, we propose a novel technique, H-Pro to drive HPO via test proxies by exploiting data hierarchies often associated with time series datasets. Since higher-level aggregated time series often show less irregularity and better predictability as compared to the lowest-level time series which can be sparse and intermittent, we optimize the hyperparameters of the lowest-level base-forecaster by leveraging the proxy forecasts for the test period generated from the forecasters at higher levels. H-Pro can be applied on any off-the-shelf machine learning model to perform HPO. We validate the efficacy of our technique with extensive empirical evaluation on five publicly available hierarchical forecasting datasets. Our approach outperforms existing state-of-the-art methods in Tourism, Wiki, and Traffic datasets, and achieves competitive result in Tourism-L dataset, without any model-specific enhancements. Moreover, our method outperforms the winning method of the M5 forecast accuracy competition.

The Koopman operator theory provides a recipe for data-driven analysis of dynamical systems. The forecasting problem addresses the future value estimation of an observation from its recent past. One common modeling approach is producing forecasts via modeling the data-generating process, where observations are a function of the state variables of the data-generating process. One major challenge to this is the lack of knowledge of the data-generating process. Delay embedding is a common approach in dynamical system analysis to approximate the state-space manifold from a few state value measurements. In this work, we propose a deep learning framework that employs delay embedding, and Koopman operator theory in the context of univariate forecasting to produce a long-term stable forecast. In this work, we empirically show the correctness of the proposed framework. Our study shows the proposed model can generalize across data-set and is capable of producing a stable long-range forecast.

Trend driven retail industries such as fashion, launch substantial new products every season. In such a scenario, an accurate demand forecast for these newly launched products is vital for efficient downstream supply chain planning like assortment planning and stock allocation. While classical time-series forecasting algorithms can be used for existing products to forecast the sales, new products do not have any historical time-series data to base the forecast on. In this paper, we propose and empirically evaluate several novel attention-based multi-modal encoder-decoder models to forecast the sales for a new product purely based on product images, any available product attributes and also external factors like holidays, events, weather, and discount. We experimentally validate our approaches on a large fashion dataset and report the improvements in achieved accuracy and enhanced model interpretability as compared to existing k-nearest neighbor based baseline approaches.

Genetic differences contribute to variations in the immune response mounted by different individuals to a pathogen. Such differential response can influence the spread of infectious disease, indicating why such diseases impact some populations more than others. Here, we study the impact of population-level genetic heterogeneity on the epidemic spread of different strains of H1N1 influenza. For a population with known HLA class-I allele frequency and for a given H1N1 viral strain, we classify individuals into sub-populations according to their level of susceptibility to infection. Our core hypothesis is that the susceptibility of a given individual to a disease such as H1N1 influenza is inversely proportional to the number of high affinity viral epitopes the individual can present. This number can be extracted from the HLA genetic profile of the individual. We use ethnicity-specific HLA class-I allele frequency data, together with genome sequences of various H1N1 viral strains, to obtain susceptibility sub-populations for 61 ethnicities and 81 viral strains isolated in 2009, as well as 85 strains isolated in other years. We incorporate these data into a multi-compartment SIR model to analyse the epidemic dynamics for these (ethnicity, viral strain) epidemic pairs. Our results show that HLA allele profiles which lead to a large spread in individual susceptibility values can act as a protective barrier against the spread of influenza. We predict that populations skewed such that a small number of highly susceptible individuals coexist with a large number of less susceptible ones, should exhibit smaller outbreaks than populations with the same average susceptibility but distributed more uniformly across individuals. Our model tracks some well-known qualitative trends of influenza spread worldwide, suggesting that HLA genetic diversity plays a crucial role in determining the spreading potential of different influenza viral strains across populations.

MOTIVATION: T-cell epitopes serve as molecular keys to initiate adaptive immune responses. Identification of T-cell epitopes is also a key step in rational vaccine design. Most available methods are driven by informatics and are critically dependent on experimentally obtained training data. Analysis of a training set from Immune Epitope Database (IEDB) for several alleles indicates that the sampling of the peptide space is extremely sparse covering a tiny fraction of the possible nonamer space, and also heavily skewed, thus restricting the range of epitope prediction. RESULTS: We present a new epitope prediction method that has four distinct computational modules: (i) structural modelling, estimating statistical pair-potentials and constraint derivation, (ii) implicit modelling and interaction profiling, (iii) feature representation and binding affinity prediction and (iv) use of graphical models to extract peptide sequence signatures to predict epitopes for HLA class I alleles. CONCLUSIONS: HLaffy is a novel and efficient epitope prediction method that predicts epitopes for any Class-1 HLA allele, by estimating the binding strengths of peptide-HLA complexes which is achieved through learning pair-potentials important for peptide binding. It relies on the strength of the mechanistic understanding of peptide-HLA recognition and provides an estimate of the total ligand space for each allele. The performance of HLaffy is seen to be superior to the currently available methods. AVAILABILITY AND IMPLEMENTATION: The method is made accessible through a webserver http://proline.biochem.iisc.ernet.in/HLaffy CONTACT: : [email protected] SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.