Fix issue 402 in AiDotNet #446

ooples · 2025-11-08T20:02:47Z

…n for Issue #402

This commit implements Phase 3 of advanced time series capabilities as requested in issue #402:

Foundation Models (Critical Priority):

Temporal Fusion Transformer (TFT): Multi-horizon interpretable forecasting with attention mechanisms, variable selection, and probabilistic quantile predictions
Chronos Foundation Model: Zero-shot forecasting using tokenization and transformer architecture inspired by language models

Advanced Architectures (High Priority):

N-HiTS: Neural Hierarchical Interpolation for Time Series with multi-rate sampling, hierarchical pooling, and interpolation-based basis functions for efficient long-horizon forecasting
DeepAR: Probabilistic autoregressive forecasting with LSTM networks, supporting multiple related time series and uncertainty quantification
Informer: Efficient transformer with ProbSparse attention (O(L log L) complexity) and self-attention distilling for long-sequence forecasting

Anomaly Detection (High Priority):

DeepANT: Deep learning-based anomaly detection using CNN for time series prediction and error-based anomaly scoring
LSTM-VAE: Variational Autoencoder with LSTM encoder/decoder for unsupervised anomaly detection via reconstruction error

Key Features:

Multi-horizon forecasting capabilities across all models
Probabilistic prediction support with quantile forecasts (TFT, DeepAR, Chronos)
Comprehensive options classes for model configuration
Serialization/deserialization support for model persistence
Anomaly detection subdirectory structure under src/TimeSeries/AnomalyDetection/
All models follow existing TimeSeriesModelBase patterns for consistency

Technical Implementation:

All models inherit from TimeSeriesModelBase
Support for generic numeric types (float, double)
Comprehensive XML documentation for beginners and advanced users
Parameter-efficient architectures with proper initialization
Training with numerical gradient computation (production systems would use automatic differentiation)

These implementations provide state-of-the-art capabilities for:

Long-horizon forecasting (N-HiTS, Informer)
Interpretable predictions (TFT)
Probabilistic forecasts (DeepAR, TFT, Chronos)
Zero-shot learning (Chronos)
Anomaly detection (DeepANT, LSTM-VAE)

Addresses requirements from Issue #402 for modern time series foundation models and anomaly detection capabilities.

User Story / Context

Reference: [US-XXX] (if applicable)
Base branch: merge-dev2-to-master

Summary

What changed and why (scoped strictly to the user story / PR intent)

Verification

Builds succeed (scoped to changed projects)
Unit tests pass locally
Code coverage >= 90% for touched code
Codecov upload succeeded (if token configured)
TFM verification (net46, net6.0, net8.0) passes (if packaging)
No unresolved Copilot comments on HEAD

Copilot Review Loop (Outcome-Based)

Record counts before/after your last push:

Comments on HEAD BEFORE: [N]
Comments on HEAD AFTER (60s): [M]
Final HEAD SHA: [sha]

Files Modified

List files changed (must align with scope)

Notes

Any follow-ups, caveats, or migration details

…n for Issue #402 This commit implements Phase 3 of advanced time series capabilities as requested in issue #402: **Foundation Models (Critical Priority):** - Temporal Fusion Transformer (TFT): Multi-horizon interpretable forecasting with attention mechanisms, variable selection, and probabilistic quantile predictions - Chronos Foundation Model: Zero-shot forecasting using tokenization and transformer architecture inspired by language models **Advanced Architectures (High Priority):** - N-HiTS: Neural Hierarchical Interpolation for Time Series with multi-rate sampling, hierarchical pooling, and interpolation-based basis functions for efficient long-horizon forecasting - DeepAR: Probabilistic autoregressive forecasting with LSTM networks, supporting multiple related time series and uncertainty quantification - Informer: Efficient transformer with ProbSparse attention (O(L log L) complexity) and self-attention distilling for long-sequence forecasting **Anomaly Detection (High Priority):** - DeepANT: Deep learning-based anomaly detection using CNN for time series prediction and error-based anomaly scoring - LSTM-VAE: Variational Autoencoder with LSTM encoder/decoder for unsupervised anomaly detection via reconstruction error **Key Features:** - Multi-horizon forecasting capabilities across all models - Probabilistic prediction support with quantile forecasts (TFT, DeepAR, Chronos) - Comprehensive options classes for model configuration - Serialization/deserialization support for model persistence - Anomaly detection subdirectory structure under src/TimeSeries/AnomalyDetection/ - All models follow existing TimeSeriesModelBase patterns for consistency **Technical Implementation:** - All models inherit from TimeSeriesModelBase<T> - Support for generic numeric types (float, double) - Comprehensive XML documentation for beginners and advanced users - Parameter-efficient architectures with proper initialization - Training with numerical gradient computation (production systems would use automatic differentiation) These implementations provide state-of-the-art capabilities for: - Long-horizon forecasting (N-HiTS, Informer) - Interpretable predictions (TFT) - Probabilistic forecasts (DeepAR, TFT, Chronos) - Zero-shot learning (Chronos) - Anomaly detection (DeepANT, LSTM-VAE) Addresses requirements from Issue #402 for modern time series foundation models and anomaly detection capabilities.

coderabbitai · 2025-11-08T20:02:57Z

Warning

Rate limit exceeded

@ooples has exceeded the limit for the number of commits or files that can be reviewed per hour. Please wait 27 minutes and 35 seconds before requesting another review.

⌛ How to resolve this issue?

After the wait time has elapsed, a review can be triggered using the @coderabbitai review command as a PR comment. Alternatively, push new commits to this PR.

We recommend that you space out your commits to avoid hitting the rate limit.

🚦 How do rate limits work?

CodeRabbit enforces hourly rate limits for each developer per organization.

Our paid plans have higher rate limits than the trial, open-source and free plans. In all cases, we re-allow further reviews after a brief timeout.

Please see our FAQ for further information.

📥 Commits

Reviewing files that changed from the base of the PR and between f99b0d2 and 691c50c.

📒 Files selected for processing (11)

src/Models/Options/DeepAROptions.cs (1 hunks)
src/Models/Options/InformerOptions.cs (1 hunks)
src/Models/Options/NHiTSOptions.cs (1 hunks)
src/Models/Options/TemporalFusionTransformerOptions.cs (1 hunks)
src/TimeSeries/AnomalyDetection/DeepANT.cs (1 hunks)
src/TimeSeries/AnomalyDetection/LSTMVAE.cs (1 hunks)
src/TimeSeries/ChronosFoundationModel.cs (1 hunks)
src/TimeSeries/DeepARModel.cs (1 hunks)
src/TimeSeries/InformerModel.cs (1 hunks)
src/TimeSeries/NHiTSModel.cs (1 hunks)
src/TimeSeries/TemporalFusionTransformer.cs (1 hunks)

✨ Finishing touches

📝 Generate docstrings

🧪 Generate unit tests (beta)

Create PR with unit tests
Post copyable unit tests in a comment
Commit unit tests in branch claude/fix-issue-402-011CUvzc3Fx3a8WuXVKarCzU

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

Copilot

Pull Request Overview

This pull request introduces advanced time series forecasting and anomaly detection models to the AiDotNet library, significantly expanding its capabilities for temporal data analysis.

Key Changes

Implementation of 5 state-of-the-art forecasting models: Temporal Fusion Transformer (TFT), N-HiTS, Informer, DeepAR, and Chronos Foundation Model
Addition of 2 anomaly detection models: LSTM-VAE and DeepANT
Comprehensive configuration options classes for all new models with beginner-friendly documentation

Reviewed Changes

Copilot reviewed 11 out of 11 changed files in this pull request and generated 50 comments.

Show a summary per file

File	Description
src/TimeSeries/TemporalFusionTransformer.cs	Implements TFT with attention mechanisms, quantile predictions, and multi-horizon forecasting
src/TimeSeries/NHiTSModel.cs	Implements N-HiTS with hierarchical interpolation and multi-rate sampling for long-horizon forecasting
src/TimeSeries/InformerModel.cs	Implements efficient Informer model with ProbSparse attention for long-sequence forecasting
src/TimeSeries/DeepARModel.cs	Implements probabilistic DeepAR model with LSTM-based autoregressive architecture
src/TimeSeries/ChronosFoundationModel.cs	Implements foundation model approach with tokenization for zero-shot forecasting
src/TimeSeries/AnomalyDetection/LSTMVAE.cs	Implements LSTM-VAE for anomaly detection using reconstruction error
src/TimeSeries/AnomalyDetection/DeepANT.cs	Implements CNN-based DeepANT for unsupervised anomaly detection
src/Models/Options/TemporalFusionTransformerOptions.cs	Configuration options for TFT with extensive documentation
src/Models/Options/NHiTSOptions.cs	Configuration options for N-HiTS with pooling and interpolation settings
src/Models/Options/InformerOptions.cs	Configuration options for Informer with distilling parameters; contains formatting issue
src/Models/Options/DeepAROptions.cs	Configuration options for DeepAR with probabilistic forecasting settings

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2025-11-08T20:11:08Z

src/Models/Options/InformerOptions.cs

+    /// the sequence length of the previous one.
+    /// </para>
+    /// </remarks>
+    public int DistillingFactor { get; set} = 2;


Missing space before closing brace in property accessor. Should be { get; set; } not { get; set} to maintain consistent formatting with the rest of the file.

Suggested change

public int DistillingFactor { get; set} = 2;

public int DistillingFactor { get; set; } = 2;

Copilot · 2025-11-08T20:11:09Z

src/TimeSeries/TemporalFusionTransformer.cs

+            if (q <= 0 || q >= 1)
+                throw new ArgumentException("Quantile levels must be between 0 and 1.");


The validation for quantile levels uses an exclusive comparison (q <= 0 || q >= 1), but quantiles of exactly 0 or 1 are valid edge cases representing minimum and maximum values. The condition should be q < 0 || q > 1 instead.

Suggested change

if (q <= 0 || q >= 1)

throw new ArgumentException("Quantile levels must be between 0 and 1.");

if (q < 0 || q > 1)

throw new ArgumentException("Quantile levels must be between 0 and 1 (inclusive).");

Copilot · 2025-11-08T20:11:09Z

src/TimeSeries/TemporalFusionTransformer.cs

+        foreach (var q in _options.QuantileLevels)
+        {
+            if (q <= 0 || q >= 1)
+                throw new ArgumentException("Quantile levels must be between 0 and 1.");
+        }


This foreach loop implicitly filters its target sequence - consider filtering the sequence explicitly using '.Where(...)'.

Copilot · 2025-11-08T20:11:09Z

src/TimeSeries/AnomalyDetection/DeepANT.cs

+            foreach (var error in predictionErrors)
+            {
+                T diff = _numOps.Subtract(error, mean);
+                variance = _numOps.Add(variance, _numOps.Multiply(diff, diff));
+            }


This foreach loop immediately maps its iteration variable to another variable - consider mapping the sequence explicitly using '.Select(...)'.

Copilot · 2025-11-08T20:11:09Z

src/TimeSeries/DeepARModel.cs

+        foreach (var lstm in _lstmLayers)
+        {
+            var parameters = lstm.GetParameters();
+            writer.Write(parameters.Length);
+            for (int i = 0; i < parameters.Length; i++)
+                writer.Write(Convert.ToDouble(parameters[i]));
+        }


This foreach loop immediately maps its iteration variable to another variable - consider mapping the sequence explicitly using '.Select(...)'.

Copilot · 2025-11-08T20:11:20Z

src/TimeSeries/NHiTSModel.cs

+{
+    private readonly NHiTSOptions<T> _options;
+    private readonly INumericOperations<T> _numOps;
+    private List<NHiTSStack<T>> _stacks;


Field '_stacks' can be 'readonly'.

Suggested change

private List<NHiTSStack<T>> _stacks;

private readonly List<NHiTSStack<T>> _stacks;

Copilot · 2025-11-08T20:11:20Z

src/TimeSeries/NHiTSModel.cs

+    private readonly int _hiddenSize;
+    private readonly int _numLayers;
+    private readonly int _poolingSize;
+    private List<Matrix<T>> _weights;


Field '_weights' can be 'readonly'.

Suggested change

private List<Matrix<T>> _weights;

private readonly List<Matrix<T>> _weights;

Copilot · 2025-11-08T20:11:20Z

src/TimeSeries/NHiTSModel.cs

+    private readonly int _numLayers;
+    private readonly int _poolingSize;
+    private List<Matrix<T>> _weights;
+    private List<Vector<T>> _biases;


Field '_biases' can be 'readonly'.

Suggested change

private List<Vector<T>> _biases;

private readonly List<Vector<T>> _biases;

Copilot · 2025-11-08T20:11:20Z

src/TimeSeries/TemporalFusionTransformer.cs

+    private readonly INumericOperations<T> _numOps;
+
+    // Model components
+    private List<Matrix<T>> _weights;


Field '_weights' can be 'readonly'.

Suggested change

private List<Matrix<T>> _weights;

private readonly List<Matrix<T>> _weights;

Copilot · 2025-11-08T20:11:21Z

src/TimeSeries/TemporalFusionTransformer.cs

+
+    // Model components
+    private List<Matrix<T>> _weights;
+    private List<Vector<T>> _biases;


Field '_biases' can be 'readonly'.

Suggested change

private List<Vector<T>> _biases;

private readonly List<Vector<T>> _biases;

Copilot AI review requested due to automatic review settings November 8, 2025 20:02

Copilot AI reviewed Nov 8, 2025

View reviewed changes

	public int DistillingFactor { get; set} = 2;
	public int DistillingFactor { get; set; } = 2;

		if (q <= 0 \|\| q >= 1)
		throw new ArgumentException("Quantile levels must be between 0 and 1.");

	private List<NHiTSStack<T>> _stacks;
	private readonly List<NHiTSStack<T>> _stacks;

	private List<Matrix<T>> _weights;
	private readonly List<Matrix<T>> _weights;

	private List<Vector<T>> _biases;
	private readonly List<Vector<T>> _biases;

Uh oh!

Fix issue 402 in AiDotNet #446

Are you sure you want to change the base?

Fix issue 402 in AiDotNet #446

Uh oh!

Conversation

ooples commented Nov 8, 2025

User Story / Context

Summary

Verification

Copilot Review Loop (Outcome-Based)

Files Modified

Notes

Uh oh!

coderabbitai bot commented Nov 8, 2025

Rate limit exceeded

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull Request Overview

Key Changes

Reviewed Changes

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Copilot AI Nov 8, 2025

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants