510_23BAI10157_23BAI10274

510_23BAI10157_23BAI10274/README.md

@@ -0,0 +1,72 @@
+# 🧠 Synapse — AI-Powered Study Planner
+
+Synapse is an intelligent, personalized study planner that uses **Reinforcement Learning (Q-Learning)** to generate optimized study roadmaps based on your strengths and weaknesses.
+
+It combines AI decision-making with a clean GUI to help you focus on what matters most.
+
+---
+
+## 🚀 Features
+
+- 🎯 Personalized Study Plan based on your current skill levels  
+- 🧠 AI-Powered Scheduling using Q-learning  
+- 📊 Adaptive Learning – improves recommendations over time  
+- 🔄 Balanced Subject Distribution (avoids repetition & burnout)  
+- 💡 Actionable Tasks for each subject (Beginner → Advanced)  
+- 🖥️ Modern GUI built with Tkinter  
+
+---
+
+## 🗂️ Project Structure
+
+.
+├── app.py
+├── brain.py
+├── logic.py
+└── q_table.json
+
+---
+
+## ⚙️ How It Works
+
+1. Input your confidence levels for each subject (0–100%)
+2. The AI identifies weak areas
+3. Generates a study sequence using reinforcement learning
+4. Displays a roadmap with actionable tasks
+
+---
+
+## 🧠 AI Approach
+
+- Uses Q-Learning
+- State: (time_slot, weakest_subject)
+- Action: next subject to study
+- Reward system encourages improvement and balance
+
+---
+
+## 🖥️ Installation & Setup
+
+### Clone the repository
+git clone https://github.com/manasvi-maheshwari/synapse-study-planner.git
+
+### Install dependencies
+pip install numpy
+
+### Run the application
+python app.py
+
+---
+
+## 🔮 Future Improvements
+
+- Data visualization
+- Web & mobile versions
+- Cloud sync
+- More customization
+
+---
+
+## 📄 License
+
+MIT License

510_23BAI10157_23BAI10274/brain.py

@@ -0,0 +1,28 @@
+import numpy as np
+import json
+import os
+
+class StudyAI:
+    def __init__(self, subjects, slots=6):
+        self.subjects = subjects
+        self.q = np.zeros((slots + 1, len(subjects), len(subjects)))
+
+    def choose(self, slot, weakest_idx, explore=True):
+        import random
+        if explore and random.random() < 0.2:
+            return random.randint(0, len(self.subjects) - 1)
+        return int(np.argmax(self.q[slot][weakest_idx]))
+
+    def learn(self, slot, weakest, choice, reward, next_slot, next_weakest):
+        cur = self.q[slot][weakest][choice]
+        best = np.max(self.q[next_slot][next_weakest])
+        self.q[slot][weakest][choice] += 0.1 * (reward + 0.9 * best - cur)
+
+    def save(self, file="q_table.json"):
+        with open(file, "w") as f: json.dump(self.q.tolist(), f)
+
+    def load(self, file="q_table.json"):
+        if os.path.exists(file):
+            with open(file) as f: self.q = np.array(json.load(f))
+            return True
+        return False

510_23BAI10157_23BAI10274/logic.py

@@ -0,0 +1,61 @@
+import random
+
+def run_simulation(ai, subjects, difficulty, scores, slots=6, explore=True):
+    current_scores = scores.copy()
+    history = []
+    last_choice = -1
+    recent_choices = []          # track last N picks for cooldown
+    subject_counts = {s: 0 for s in subjects}
+    n = len(subjects)
+
+    # Ensure slots >= number of subjects so every subject gets at least one slot
+    slots = max(slots, n)
+
+    for slot in range(slots):
+        weakest_idx = min(range(n), key=lambda i: current_scores[subjects[i]])
+
+        # --- Cooldown mask: block subjects picked in last (n-1) turns ---
+        cooldown = max(1, n - 1)
+        blocked = set(recent_choices[-cooldown:]) if len(subjects) > 1 else set()
+
+        # Force-include any subject that has had zero slots so far
+        unvisited = [i for i in range(n) if subject_counts[subjects[i]] == 0]
+        if unvisited:
+            # Pick the weakest unvisited subject
+            choice = min(unvisited, key=lambda i: current_scores[subjects[i]])
+        else:
+            # Let AI choose, but mask out recently used subjects
+            raw_choice = ai.choose(slot, weakest_idx, explore)
+            if raw_choice in blocked:
+                # Fall back to weakest subject not on cooldown
+                available = [i for i in range(n) if i not in blocked]
+                if not available:
+                    available = list(range(n))   # safety valve
+                choice = min(available, key=lambda i: current_scores[subjects[i]])
+            else:
+                choice = raw_choice
+
+        subj = subjects[choice]
+        subject_counts[subj] += 1
+
+        # --- Gain: diminishing returns as score rises ---
+        gap = 1.0 - current_scores[subj]
+        gain = gap * difficulty[subj] * random.uniform(0.12, 0.22)
+        current_scores[subj] = min(1.0, current_scores[subj] + gain)
+
+        # --- Reward shaping ---
+        reward = gain * 20
+        if choice == weakest_idx:
+            reward += 5                          # good: tackled weakest
+        if choice == last_choice:
+            reward -= 15                         # strong repetition penalty
+        if subject_counts[subj] > slots // n + 1:
+            reward -= 10                         # penalise over-concentration
+
+        ai.learn(slot, weakest_idx, choice, reward, slot + 1, weakest_idx)
+        history.append({"subj": subj, "score": current_scores[subj], "gain": gain})
+
+        recent_choices.append(choice)
+        last_choice = choice
+
+    return history

README.md

@@ -0,0 +1,72 @@
+# 🧠 Synapse — AI-Powered Study Planner
+
+Synapse is an intelligent, personalized study planner that uses **Reinforcement Learning (Q-Learning)** to generate optimized study roadmaps based on your strengths and weaknesses.
+
+It combines AI decision-making with a clean GUI to help you focus on what matters most.
+
+---
+
+## 🚀 Features
+
+- 🎯 Personalized Study Plan based on your current skill levels  
+- 🧠 AI-Powered Scheduling using Q-learning  
+- 📊 Adaptive Learning – improves recommendations over time  
+- 🔄 Balanced Subject Distribution (avoids repetition & burnout)  
+- 💡 Actionable Tasks for each subject (Beginner → Advanced)  
+- 🖥️ Modern GUI built with Tkinter  
+
+---
+
+## 🗂️ Project Structure
+
+.
+├── app.py
+├── brain.py
+├── logic.py
+└── q_table.json
+
+---
+
+## ⚙️ How It Works
+
+1. Input your confidence levels for each subject (0–100%)
+2. The AI identifies weak areas
+3. Generates a study sequence using reinforcement learning
+4. Displays a roadmap with actionable tasks
+
+---
+
+## 🧠 AI Approach
+
+- Uses Q-Learning
+- State: (time_slot, weakest_subject)
+- Action: next subject to study
+- Reward system encourages improvement and balance
+
+---
+
+## 🖥️ Installation & Setup
+
+### Clone the repository
+git clone https://github.com/manasvi-maheshwari/synapse-study-planner.git
+
+### Install dependencies
+pip install numpy
+
+### Run the application
+python app.py
+
+---
+
+## 🔮 Future Improvements
+
+- Data visualization
+- Web & mobile versions
+- Cloud sync
+- More customization
+
+---
+
+## 📄 License
+
+MIT License