File size: 20,105 Bytes
23804b3 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 |
"""
Meta-Cognitive Capabilities for Cyber-LLM
Self-reflection, adaptation, and cognitive load management
Author: Muzan Sano <[email protected]>
"""
import asyncio
import json
import logging
import numpy as np
from datetime import datetime, timedelta
from typing import Dict, List, Any, Optional, Tuple, Union
from dataclasses import dataclass, field
from enum import Enum
import torch
import torch.nn as nn
from collections import deque
from ..utils.logging_system import CyberLLMLogger, CyberLLMError, ErrorCategory
from ..memory.persistent_memory import PersistentMemoryManager
from ..memory.strategic_planning import StrategicPlanningEngine
class CognitiveState(Enum):
"""Cognitive processing states"""
OPTIMAL = "optimal"
MODERATE_LOAD = "moderate_load"
HIGH_LOAD = "high_load"
OVERLOADED = "overloaded"
RECOVERING = "recovering"
class AdaptationStrategy(Enum):
"""Learning adaptation strategies"""
AGGRESSIVE = "aggressive"
MODERATE = "moderate"
CONSERVATIVE = "conservative"
CAUTIOUS = "cautious"
@dataclass
class CognitiveMetrics:
"""Cognitive performance metrics"""
timestamp: datetime
# Performance metrics
task_completion_rate: float
accuracy_score: float
response_time: float
resource_utilization: float
# Cognitive load indicators
attention_fragmentation: float # 0-1, higher = more fragmented
working_memory_usage: float # 0-1, percentage used
processing_complexity: float # 0-1, task complexity measure
# Adaptation metrics
learning_rate: float
confidence_level: float
adaptation_success_rate: float
# Error metrics
error_count: int
critical_errors: int
recovery_time: Optional[float] = None
@dataclass
class SelfReflectionResult:
"""Results from self-reflection analysis"""
reflection_id: str
timestamp: datetime
# Performance assessment
strengths: List[str]
weaknesses: List[str]
improvement_areas: List[str]
# Strategy effectiveness
effective_strategies: List[str]
ineffective_strategies: List[str]
recommended_changes: List[str]
# Cognitive insights
cognitive_patterns: Dict[str, Any]
load_management_insights: List[str]
attention_allocation_insights: List[str]
# Action items
immediate_adjustments: List[str]
medium_term_goals: List[str]
long_term_objectives: List[str]
class MetaCognitiveEngine:
"""Advanced meta-cognitive capabilities for self-reflection and adaptation"""
def __init__(self,
memory_manager: PersistentMemoryManager,
strategic_planner: StrategicPlanningEngine,
logger: Optional[CyberLLMLogger] = None):
self.memory_manager = memory_manager
self.strategic_planner = strategic_planner
self.logger = logger or CyberLLMLogger(name="meta_cognitive")
# Cognitive state tracking
self.current_state = CognitiveState.OPTIMAL
self.state_history = deque(maxlen=1000)
self.cognitive_metrics = deque(maxlen=10000)
# Self-reflection system
self.reflection_history = {}
self.performance_baselines = {}
self.adaptation_strategies = {}
# Cognitive load management
self.attention_allocator = AttentionAllocator()
self.cognitive_load_monitor = CognitiveLoadMonitor()
# Learning optimization
self.learning_rate_optimizer = LearningRateOptimizer()
self.strategy_evaluator = StrategyEvaluator()
# Neural networks for meta-learning
self.performance_predictor = self._build_performance_predictor()
self.strategy_selector = self._build_strategy_selector()
self.logger.info("Meta-Cognitive Engine initialized")
async def conduct_self_reflection(self,
time_period: timedelta = timedelta(hours=1)) -> SelfReflectionResult:
"""Conduct comprehensive self-reflection analysis"""
reflection_id = f"reflection_{datetime.now().strftime('%Y%m%d_%H%M%S')}"
try:
self.logger.info("Starting self-reflection analysis", reflection_id=reflection_id)
# Gather performance data
end_time = datetime.now()
start_time = end_time - time_period
performance_data = await self._gather_performance_data(start_time, end_time)
cognitive_data = await self._gather_cognitive_data(start_time, end_time)
strategy_data = await self._gather_strategy_data(start_time, end_time)
# Analyze strengths and weaknesses
strengths, weaknesses = await self._analyze_performance_patterns(performance_data)
# Evaluate strategy effectiveness
effective_strategies, ineffective_strategies = await self._evaluate_strategies(strategy_data)
# Generate insights
cognitive_patterns = await self._analyze_cognitive_patterns(cognitive_data)
load_insights = await self._analyze_load_management(cognitive_data)
attention_insights = await self._analyze_attention_allocation(cognitive_data)
# Generate recommendations
immediate_adjustments = await self._generate_immediate_adjustments(
weaknesses, ineffective_strategies, cognitive_patterns)
medium_term_goals = await self._generate_medium_term_goals(
strengths, weaknesses, cognitive_patterns)
long_term_objectives = await self._generate_long_term_objectives(
performance_data, cognitive_patterns)
# Create reflection result
reflection_result = SelfReflectionResult(
reflection_id=reflection_id,
timestamp=datetime.now(),
strengths=strengths,
weaknesses=weaknesses,
improvement_areas=list(set(weaknesses + [adj.split(':')[0] for adj in immediate_adjustments])),
effective_strategies=effective_strategies,
ineffective_strategies=ineffective_strategies,
recommended_changes=immediate_adjustments + medium_term_goals,
cognitive_patterns=cognitive_patterns,
load_management_insights=load_insights,
attention_allocation_insights=attention_insights,
immediate_adjustments=immediate_adjustments,
medium_term_goals=medium_term_goals,
long_term_objectives=long_term_objectives
)
# Store reflection result
self.reflection_history[reflection_id] = reflection_result
# Store in persistent memory
await self.memory_manager.store_reasoning_chain(
chain_id=f"self_reflection_{reflection_id}",
steps=[
f"Analyzed performance over {time_period}",
f"Identified {len(strengths)} strengths and {len(weaknesses)} weaknesses",
f"Evaluated {len(effective_strategies)} effective strategies",
f"Generated {len(immediate_adjustments)} immediate adjustments"
],
conclusion=f"Self-reflection completed with actionable insights",
confidence=0.85,
metadata={
"reflection_type": "comprehensive_analysis",
"time_period": str(time_period),
"performance_score": np.mean([m.accuracy_score for m in self.cognitive_metrics if m.timestamp >= start_time])
}
)
self.logger.info("Self-reflection analysis completed",
reflection_id=reflection_id,
strengths_count=len(strengths),
weaknesses_count=len(weaknesses),
recommendations_count=len(immediate_adjustments))
return reflection_result
except Exception as e:
self.logger.error("Self-reflection analysis failed", error=str(e))
raise CyberLLMError("Self-reflection failed", ErrorCategory.COGNITIVE_ERROR)
async def optimize_learning_rate(self,
recent_performance: List[float],
task_complexity: float) -> float:
"""Optimize learning rate based on recent performance and task complexity"""
try:
# Analyze performance trends
performance_trend = self._calculate_performance_trend(recent_performance)
performance_variance = np.var(recent_performance)
# Current learning rate
current_lr = self.learning_rate_optimizer.get_current_rate()
# Adaptation strategy based on performance
if performance_trend > 0.1 and performance_variance < 0.05:
# Good performance, stable -> slightly increase learning rate
adaptation_factor = 1.1
strategy = AdaptationStrategy.AGGRESSIVE
elif performance_trend > 0.05:
# Moderate improvement -> maintain or slight increase
adaptation_factor = 1.05
strategy = AdaptationStrategy.MODERATE
elif performance_trend < -0.1 or performance_variance > 0.2:
# Poor performance or high variance -> decrease learning rate
adaptation_factor = 0.8
strategy = AdaptationStrategy.CAUTIOUS
else:
# Stable performance -> minor adjustment based on complexity
adaptation_factor = 1.0 - (task_complexity - 0.5) * 0.1
strategy = AdaptationStrategy.CONSERVATIVE
# Apply complexity adjustment
complexity_factor = 1.0 - (task_complexity * 0.3)
final_factor = adaptation_factor * complexity_factor
# Calculate new learning rate
new_lr = current_lr * final_factor
new_lr = np.clip(new_lr, 0.0001, 0.1) # Keep within reasonable bounds
# Update learning rate optimizer
self.learning_rate_optimizer.update_rate(new_lr, strategy)
self.logger.info("Learning rate optimized",
old_rate=current_lr,
new_rate=new_lr,
strategy=strategy.value,
performance_trend=performance_trend)
return new_lr
except Exception as e:
self.logger.error("Learning rate optimization failed", error=str(e))
return self.learning_rate_optimizer.get_current_rate()
async def manage_cognitive_load(self,
current_tasks: List[Dict[str, Any]],
available_resources: Dict[str, float]) -> Dict[str, Any]:
"""Manage cognitive load and optimize task allocation"""
try:
# Calculate current cognitive load
current_load = await self._calculate_cognitive_load(current_tasks)
# Determine cognitive state
new_state = self._determine_cognitive_state(current_load, available_resources)
# Update state if changed
if new_state != self.current_state:
self.logger.info("Cognitive state changed",
old_state=self.current_state.value,
new_state=new_state.value)
self.current_state = new_state
self.state_history.append((datetime.now(), new_state))
# Generate load management strategy
management_strategy = await self._generate_load_management_strategy(
current_load, new_state, current_tasks, available_resources)
# Apply attention allocation optimization
attention_allocation = await self.attention_allocator.optimize_allocation(
current_tasks, available_resources, new_state)
# Generate recommendations
recommendations = await self._generate_load_management_recommendations(
current_load, new_state, management_strategy)
result = {
"cognitive_state": new_state.value,
"cognitive_load": current_load,
"management_strategy": management_strategy,
"attention_allocation": attention_allocation,
"recommendations": recommendations,
"resource_adjustments": await self._calculate_resource_adjustments(
new_state, available_resources)
}
self.logger.info("Cognitive load management completed",
state=new_state.value,
load=current_load,
recommendations_count=len(recommendations))
return result
except Exception as e:
self.logger.error("Cognitive load management failed", error=str(e))
return {"error": str(e)}
def _build_performance_predictor(self) -> nn.Module:
"""Build neural network for performance prediction"""
class PerformancePredictor(nn.Module):
def __init__(self):
super().__init__()
self.fc1 = nn.Linear(20, 64) # Input: various metrics
self.fc2 = nn.Linear(64, 32)
self.fc3 = nn.Linear(32, 16)
self.fc4 = nn.Linear(16, 1) # Output: predicted performance
self.dropout = nn.Dropout(0.2)
def forward(self, x):
x = torch.relu(self.fc1(x))
x = self.dropout(x)
x = torch.relu(self.fc2(x))
x = self.dropout(x)
x = torch.relu(self.fc3(x))
x = torch.sigmoid(self.fc4(x))
return x
return PerformancePredictor()
def _build_strategy_selector(self) -> nn.Module:
"""Build neural network for strategy selection"""
class StrategySelector(nn.Module):
def __init__(self):
super().__init__()
self.fc1 = nn.Linear(15, 48) # Input: context features
self.fc2 = nn.Linear(48, 24)
self.fc3 = nn.Linear(24, 8) # Output: strategy probabilities
self.dropout = nn.Dropout(0.15)
def forward(self, x):
x = torch.relu(self.fc1(x))
x = self.dropout(x)
x = torch.relu(self.fc2(x))
x = torch.softmax(self.fc3(x), dim=-1)
return x
return StrategySelector()
class AttentionAllocator:
"""Manages dynamic attention allocation across tasks"""
def __init__(self):
self.attention_weights = {}
self.priority_scores = {}
self.allocation_history = deque(maxlen=1000)
async def optimize_allocation(self,
tasks: List[Dict[str, Any]],
resources: Dict[str, float],
cognitive_state: CognitiveState) -> Dict[str, float]:
"""Optimize attention allocation across tasks"""
# Calculate base priority scores
for task in tasks:
task_id = task.get('id', str(hash(str(task))))
priority = task.get('priority', 0.5)
complexity = task.get('complexity', 0.5)
deadline_pressure = task.get('deadline_pressure', 0.0)
# Adjust priority based on cognitive state
state_multiplier = {
CognitiveState.OPTIMAL: 1.0,
CognitiveState.MODERATE_LOAD: 0.9,
CognitiveState.HIGH_LOAD: 0.7,
CognitiveState.OVERLOADED: 0.5,
CognitiveState.RECOVERING: 0.6
}.get(cognitive_state, 1.0)
adjusted_priority = (priority * 0.4 +
deadline_pressure * 0.4 +
(1.0 - complexity) * 0.2) * state_multiplier
self.priority_scores[task_id] = adjusted_priority
# Normalize allocation
total_priority = sum(self.priority_scores.values())
if total_priority > 0:
allocation = {task_id: score / total_priority
for task_id, score in self.priority_scores.items()}
else:
# Equal allocation if no priorities
equal_weight = 1.0 / len(tasks) if tasks else 0.0
allocation = {task.get('id', str(i)): equal_weight
for i, task in enumerate(tasks)}
# Store allocation history
self.allocation_history.append((datetime.now(), allocation))
return allocation
class CognitiveLoadMonitor:
"""Monitors and analyzes cognitive load patterns"""
def __init__(self):
self.load_history = deque(maxlen=10000)
self.load_patterns = {}
def calculate_load(self,
active_tasks: int,
task_complexity: float,
resource_usage: float,
error_rate: float) -> float:
"""Calculate current cognitive load"""
# Base load from task count (logarithmic scaling)
task_load = min(np.log(active_tasks + 1) / np.log(10), 1.0)
# Complexity contribution
complexity_load = task_complexity * 0.3
# Resource pressure
resource_load = resource_usage * 0.25
# Error pressure (exponential)
error_load = min(error_rate ** 0.5, 1.0) * 0.2
total_load = task_load + complexity_load + resource_load + error_load
# Store in history
self.load_history.append((datetime.now(), total_load))
return min(total_load, 1.0)
class LearningRateOptimizer:
"""Optimizes learning rates based on performance feedback"""
def __init__(self, initial_rate: float = 0.001):
self.current_rate = initial_rate
self.rate_history = deque(maxlen=1000)
self.performance_history = deque(maxlen=1000)
self.strategy_effectiveness = {}
def get_current_rate(self) -> float:
return self.current_rate
def update_rate(self, new_rate: float, strategy: AdaptationStrategy):
self.rate_history.append((datetime.now(), self.current_rate, new_rate, strategy))
self.current_rate = new_rate
class StrategyEvaluator:
"""Evaluates effectiveness of different strategies"""
def __init__(self):
self.strategy_outcomes = {}
self.strategy_scores = {}
def record_strategy_outcome(self, strategy: str, outcome_score: float):
if strategy not in self.strategy_outcomes:
self.strategy_outcomes[strategy] = deque(maxlen=100)
self.strategy_outcomes[strategy].append((datetime.now(), outcome_score))
# Update average score
scores = [score for _, score in self.strategy_outcomes[strategy]]
self.strategy_scores[strategy] = np.mean(scores)
# Factory function
def create_meta_cognitive_engine(memory_manager: PersistentMemoryManager,
strategic_planner: StrategicPlanningEngine,
**kwargs) -> MetaCognitiveEngine:
"""Create meta-cognitive engine"""
return MetaCognitiveEngine(memory_manager, strategic_planner, **kwargs)
|