Upload 19 files

.gitattributes

@@ -33,3 +33,5 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 *.zip filter=lfs diff=lfs merge=lfs -text
 *.zst filter=lfs diff=lfs merge=lfs -text
 *tfevents* filter=lfs diff=lfs merge=lfs -text
+Neuro[[:space:]]Symbolic[[:space:]]AI[[:space:]]-[[:space:]]Forecasting[[:space:]]Trend[[:space:]]Predictor.docx filter=lfs diff=lfs merge=lfs -text
+SymbolicSeem-NeuroAI-LSTM-Stocks-Forecasting-Report.pdf filter=lfs diff=lfs merge=lfs -text

LICENSE

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024 Sanaullah Tareen
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

Neuro Symbolic AI - Forecasting Trend Predictor.docx

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:51e8e7a3aff51e237bef1a381972c45b556333960b19b2f32e4ebb508afacc59
+size 213346

README.md

@@ -1,17 +1,53 @@
+# Neuro-Symbolic AI Forecasting: The Symbolic Seam
+
+## What It Does
+Combines neural forecasting (LSTM) with Knowledge Graph constraints via Quadratic Programming. Ensures predictions respect real-world limits — even during crises.
+
+## Key Results
+| Condition | Neural Baseline | Symbolic Seam |
+|-----------|---|---|
+| Normal markets | RMSE 3.698 | RMSE 3.695 (identical) |
+| Black Swan crisis | 95.33% violations | 6.00% violations ✅ |
+| Violations prevented | — | **938 / 1,050** |
+
+## How It Works
+1. **System 1 (LSTM)**: Predicts prices from historical data
+2. **System 2 (KG)**: Defines operational bounds (min/max price, daily change limit)
+3. **Symbolic Seam**: Projects forecast into constraint space using QP solver
+
+## Quick Start
+```bash
+pip install yfinance cvxpy torch pandas scikit-learn -q
+jupyter notebook Neuro_Symbolic_Forecasting_Kaggle.ipynb
+```
+
+## Dataset
+- **Source**: Yahoo Finance (yfinance)
+- **Stocks**: 50 S&P 500 stocks
+- **Period**: Jan 2020 – Jan 2024
+- **Sequences**: 48,800 (30-day windows)
+
+## Model
+- LSTM: 64 hidden, 2 layers
+- Training: 15 epochs, batch 64
+- Best val loss: 0.000539
+
+## Key Findings
+- **Normal**: Zero accuracy penalty (3.698 → 3.695 RMSE)
+- **Crisis**: 89.33pp improvement in constraint compliance (95.33% → 6.00% violations)
+- **Per-stock**: High-volatility stocks improve most (CRM +5.2%, DHR +1.7%)
+
+## Why This Matters
+Neural models fail when rules change (tariffs, limits, regulations). Symbolic Seam enforces hard constraints at inference time without retraining.
+
+## Files
+- `Neuro_Symbolic_Forecasting_Kaggle.ipynb` — Single notebook, all stages
+- `IEEE_Symbolic_Seam_Full.docx` — Full research paper
+- `models_backup_50stock.zip` — Trained model + config
+
+## Author
+Sanaullah Tareen | PAF-IAST | [@SanaullahTareen](https://github.com/SanaullahTareen)
+
 ---
-license: mit
-datasets:
-- defeatbeta/yahoo-finance-data
-language:
-- en
-metrics:
-- r_squared
-pipeline_tag: graph-ml
-tags:
-- graph-machine-learning
-- lstm
-- neuro-ai
-- symbolic-seen
-- unpredictable-market-trend
-- stock-price-prediction
----
+
+Full details in IEEE paper or Kaggle notebook.

SymbolicSeem-NeuroAI-LSTM-Stocks-Forecasting-Report.pdf

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:68437c396feca66b4bb5b7a232b08e70ea958eb8c517d8f5ec5a73a1ba18a00f
+size 294554

configs/model_config.yaml

@@ -0,0 +1,50 @@
+# Model Configuration for Neuro-Symbolic Forecasting
+# System 1: Neural Perception Layer (LSTM)
+
+model:
+  type: "LSTM"
+  shared: true  # One model, all 50 stocks
+
+lstm:
+  input_size: 1
+  hidden_size: 64
+  num_layers: 2
+  dropout: 0.2
+  batch_first: true
+
+output_layer:
+  type: "Linear"
+  output_size: 1
+
+training:
+  optimizer: "Adam"
+  learning_rate: 0.001
+  weight_decay: 0.0
+  epochs: 15
+  batch_size: 64
+  early_stopping_patience: 3
+
+data:
+  sequence_length: 30  # 30-day look-back window
+  forecast_horizon: 1   # Predict next 1 day
+  train_split: 0.70
+  val_split: 0.15
+  test_split: 0.15
+  normalization: "MinMaxScaler"
+  scaler_range: [0, 1]
+
+hardware:
+  device: "cuda"
+  multi_gpu: true
+  gpus: 2
+  mixed_precision: false
+
+paths:
+  model_save: "models/neural_checkpoint.pth"
+  scaler_save: "models/scaler.pkl"
+  best_model: "models/multi_stock_model.pth"
+
+monitoring:
+  log_frequency: 5  # Log every 5 epochs
+  save_best: true
+  best_metric: "val_loss"

configs/rules_config.yaml

@@ -0,0 +1,58 @@
+# Knowledge Graph Configuration for Symbolic Seam
+# System 2: Symbolic Logic Layer (Constraints)
+
+constraints:
+  price_bounds:
+    type: "relative_to_mean"
+    min_multiplier: 0.95   # 5% below mean
+    max_multiplier: 1.05   # 5% above mean
+    description: "Price cannot deviate >5% from historical mean"
+
+  rate_of_change:
+    type: "daily_change_limit"
+    max_daily_change_pct: 0.05  # 5% max change per day
+    description: "Circuit breaker: limit 5% move in single day"
+
+  market_hours:
+    type: "trading_hours"
+    enabled: false
+    market_open: "09:30"
+    market_close: "16:00"
+    timezone: "America/New_York"
+
+  trading_halts:
+    type: "blackout_dates"
+    enabled: false
+    holidays: ["2024-01-01", "2024-07-04", "2024-12-25"]
+    description: "NYSE closed on holidays"
+
+solver:
+  type: "QP"
+  backend: "OSQP"
+  timeout_ms: 2
+  verbose: false
+  max_iterations: 100
+
+constraint_enforcement:
+  method: "hard"  # Hard constraints (feasibility guaranteed)
+  fallback: "clamp"  # If QP fails, clamp to bounds
+  projection_type: "l2"  # L2 norm minimization (closest point)
+
+paths:
+  kg_save: "models/kg_rules_50.pkl"
+  bounds_save: "models/bounds.json"
+
+stocks:
+  count: 50
+  list: ["AAPL", "MSFT", "GOOGL", "AMZN", "TSLA", "JPM", "V", "JNJ", 
+         "WMT", "XOM", "NVDA", "META", "PG", "KO", "DIS", "BAC", 
+         "HD", "MA", "PFE", "CSCO", "INTC", "VZ", "ADBE", "NFLX", 
+         "CRM", "ABT", "T", "MRK", "PEP", "AVGO", "COST", "TMO", 
+         "ACN", "NKE", "MCD", "LLY", "DHR", "TXN", "NEE", "UPS", 
+         "PM", "QCOM", "HON", "UNH", "LOW", "IBM", "GE", "CAT", "BA", "AMD"]
+
+stress_test:
+  black_swan_enabled: true
+  shock_magnitude: 0.50  # 50% price spike
+  test_period: "2020-03-01:2020-03-31"  # COVID crash
+  measure_violations: true

gitignore

@@ -0,0 +1,42 @@
+# Python
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+env/
+venv/
+ENV/
+build/
+dist/
+*.egg-info/
+*.egg
+
+# Jupyter
+.ipynb_checkpoints/
+*.ipynb_checkpoints
+
+# Data
+data/raw/*.csv
+data/raw/*.json
+data/processed/*.npy
+data/processed/*.pkl
+
+# Models & Results
+models/*.pth
+models/*.pkl
+results/metrics/*.csv
+results/plots/*.png
+results/predictions/*.csv
+
+# IDE
+.vscode/
+.idea/
+*.swp
+
+# OS
+.DS_Store
+Thumbs.db
+
+# Logs
+*.log

models_backup_50stocks.zip

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a45173e223a2fbe0bc17af50deb89cc8c40b780f2e7dc2f81284dc20d93cdd66
+size 758692

models_backup_50stocks/kg_rules_50.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1e482732a420519571370f1629831301a41aab426ae9ee5c2e3e1802207c600c
+size 2210

models_backup_50stocks/model_50stock.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:dd7143078d05912a6451b53adc79b94ca19bd3be3e0ef940eefb4a2e247ad3cc
+size 204951

models_backup_50stocks/model_50stocks.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:448752e5b0f17ce5b06c52c9a5784635a1809b12c0e55c61765e3d3553a5840a
+size 204960

models_backup_50stocks/multi_config.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4d15bcc04d0afd5176559372aa82b76215ecf8113fd1ccb53d01feacd3f9e000
+size 139

models_backup_50stocks/multi_kg_rules.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:de9a1ab37c4c2d87f77f63f2bfc8ae2c72405b5ce39dea74d1cab42be671ec03
+size 706

models_backup_50stocks/multi_scalers.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:60a17f08844a5c7c181f28d42a3137f15dfcc3e52a2118cec86dcb405bcd03a8
+size 3882

models_backup_50stocks/multi_stock_model.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:a68fc384dae0397859a2ee23e4d05acaf4ba5cc80b4cef84314222943beab83c
+size 204987

models_backup_50stocks/neural_checkpoint.pth

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e66972677490a0f292117f5df87d84c9704d70eea9eee6c8dfe2e2ff57433636
+size 204987

models_backup_50stocks/scalers_50.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4222911e5e4a9c7d41c533cab3f2bc5288b0fe0b436286d86a3597b09a5185b9
+size 12246

requirements.txt

@@ -0,0 +1,104 @@
+# Neuro-Symbolic AI Forecasting - Requirements
+
+# ============================================
+# Python Version
+# ============================================
+# Python 3.9+ required
+# Tested on: Python 3.10, 3.11
+
+# ============================================
+# Core Dependencies
+# ============================================
+
+# Deep Learning Framework
+torch>=2.0.0           # PyTorch for neural network and GPU support
+torchvision>=0.15.0    # For data loading utilities
+
+# Data Processing & Analysis
+numpy>=1.21.0          # Numerical computing
+pandas>=1.3.0          # Data manipulation and time series handling
+scikit-learn>=1.0.0    # Machine learning utilities (MinMaxScaler)
+
+# Financial Data
+yfinance>=0.2.28       # Download stock data from Yahoo Finance
+
+# Optimization (Symbolic Layer)
+cvxpy>=1.3.0           # Convex optimization for symbolic seam projection
+osqp>=0.6.2            # OSQP solver (backend for cvxpy)
+
+# Visualization
+matplotlib>=3.5.0      # Plotting and visualization
+
+# ============================================
+# Optional Dependencies
+# ============================================
+
+# Jupyter Notebooks
+jupyter>=1.0.0         # Jupyter notebook support
+ipython>=8.0.0         # Enhanced interactive Python
+
+# Development & Testing
+pytest>=7.0.0          # Unit testing framework
+black>=22.0.0          # Code formatting
+pylint>=2.12.0         # Code linting
+
+# ============================================
+# GPU/CPU Requirements
+# ============================================
+
+# GPU Support (Optional but Recommended)
+# For NVIDIA GPU: Install CUDA 11.8 or 12.1
+# Command: pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
+
+# For CPU-only (slower training, 10-20x slower on large datasets):
+# Already included in default torch installation
+
+# ============================================
+# Hardware Specifications
+# ============================================
+
+# Minimum (CPU-only):
+# - Processor: 4 cores
+# - RAM: 8 GB
+# - Storage: 2 GB
+# - Time to train: ~30 minutes per epoch
+
+# Recommended (GPU):
+# - GPU: NVIDIA with 4GB+ VRAM (tested on T4, V100)
+# - Processor: 8 cores
+# - RAM: 16 GB
+# - Storage: 5 GB
+# - Time to train: ~1-2 minutes per epoch
+
+# ============================================
+# Installation Instructions
+# ============================================
+
+# 1. Create virtual environment
+# python -m venv neuro-symbolic-env
+# .\neuro-symbolic-env\Scripts\activate  # Windows
+# source neuro-symbolic-env/bin/activate # Linux/Mac
+
+# 2. Install basic requirements
+# pip install -r requirements.txt
+
+# 3. For GPU support (NVIDIA only)
+# pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121
+
+# 4. Verify installation
+# python -c "import torch; print(torch.__version__); print(torch.cuda.is_available())"
+
+# ============================================
+# Version Pinning (Exact Versions Used)
+# ============================================
+
+# Uncomment below to use exact versions that were tested:
+# torch==2.1.0
+# torchvision==0.16.0
+# numpy==1.24.3
+# pandas==2.0.3
+# scikit-learn==1.3.0
+# yfinance==0.2.32
+# cvxpy==1.3.2
+# matplotlib==3.7.2
+# jupyter==1.0.0