sinovels Chapter 1436 Intuition
Chapter 1436 Intuition
Chapter 1436 Intuition
At seven o'clock sharp, Yang Ping pushed open the door and came in.
He didn't ask "Is it done?", but simply walked behind Tang Shun and looked at the screen. Tang Shun's fingers typed the last line of commands on the keyboard and then pressed Enter.
A green message popped up on the screen: "Prediction accuracy 94.7%, real-time latency 12ms, passed."
“The accuracy loss is 5.3%,” Tang Shun’s voice was hoarse, “0.3% more than you requested.”
Yang Ping didn't say anything, but simply reached out and placed his hand on Tang Shun's shoulder.
“That’s enough,” Yang Ping said. “0.3% isn’t an error for me.”
Tang Shun's back relaxed, like a bowstring that had been stretched for too long finally releasing its tension. He leaned back in his chair, closed his eyes, and his Adam's apple bobbed.
Yang Ping chatted with Tang Shun for a while, then went to the animal room, where Fritz was giving M7 his morning care. Seeing Yang Ping, Fritz didn't stand up, but simply nodded and continued his work. After three months, he had grown accustomed to Yang Ping's comings and goings. This man required no small talk, no formalities; wherever he was, work continued.
"How was last night?" Yang Ping asked.
“It woke up once at three,” Fritz said. “I turned on the nightlight, it glanced at me, and then went back to sleep.”
"No unusual behavior?"
“No,” Fritz paused, “but it dreamed.”
"dream?"
“The hind legs are moving,” Fritz gestured, “like running. I recorded a video, it’s in your shared folder.”
Yang Ping crouched down. M7's eyes were half-open, its pupils reflecting his image. It saw Yang Ping, its ears twitched, but it didn't reach out. It was still in a state between sleep and wakefulness, unable to distinguish between reality and dreams.
"What did it dream about?" Yang Ping asked softly, as if asking M7, or perhaps asking himself.
“The forest,” Fritz said, “where it used to live. I’ve seen its file; it was born in a nature reserve in Yunnan and was captured at the age of three to be used as an experimental animal. It remembers the trees, the wind, and the way sunlight filtered through the leaves.”
Yang Ping was silent for a moment, then he reached out and gently touched M7's forehead. M7's fur was soft and carried the warmth unique to animals.
“We will get it back,” he said. “Not now, but someday.”
Fritz looked at him, something flickering in his eyes. He wanted to say "thank you," but Yang Ping had already stood up, the hem of his white coat brushing against the cage bars with a slight rustling sound.
“Where is Professor Weber?” Yang Ping asked.
“In the lab,” Fritz said, “he arrived at six o’clock, saying he wanted to test a new cooling solution.”
“Let him rest,” Yang Ping said. “There’s no need to rush today.”
“I said,” Fritz said, bowing his head, “he said, ‘What Professor Yang says in a day, is a day.’”
Yang Ping paused for a moment. He didn't turn around, but simply said "I understand," and continued walking forward.
Weber wasn't rushing to finish his work. He was waiting for Yang Ping, waiting for the moment he could report back. In Germany, he never had to wait for anyone. But here, he had learned to wait. This waiting wasn't out of humility, but out of respect. Yang Ping's time was more precious than he had imagined; every minute and second was fragmented, allocated to different people and different tasks.
"Professor Weber," Yang Ping said, pushing the door open, "how are the heat sinks?"
"Running continuously for eight hours, the temperature fluctuation did not exceed one degree Celsius," Weber stood up and handed over the test records. "Better than expected."
Yang Ping took the record, didn't look at it, and put it directly on the table.
"How long did you sleep last night?" he asked.
Weber paused for a moment. This question was outside his preparedness. He had prepared technical parameters, comparative data, and the next steps, but he hadn't prepared for "how long he slept."
Weber said, "Four hours is enough."
Yang Ping said, "That's not enough. You're seventy-eight, not twenty-eight. Go home and catch up on your sleep today, then come back tomorrow."
"But……"
“This is an order,” Yang Ping said calmly but firmly, “not something to be discussed.”
Weber looked at him for a long time, and a smile appeared on his lips.
“Professor Yang,” he said, “do you know that in Germany, no one has ever dared to order me around?”
“I know,” Yang Ping said.
He walked to the testing platform, picked up the heatsink, and examined it against the light. The intricate patterns shimmered under the light, like a work of art.
“You did a great job, even better than I expected.”
Yang Ping never expected that Weber, who was almost eighty years old and a top expert in the field of stem cells, would actually know how to do this kind of thing.
Weber's eyes reddened slightly. He recalled that back in Germany, his assistants would produce excellent work, but he rarely said "excellent." He was used to being critical, used to finding fault, used to using high standards to pressure others, and also to pressure himself.
Weber remained silent for a long time, then he extended his hand and shook hands with Yang Ping.
At 3 p.m. sharp, Tang Shun appeared in Yang Ping’s office.
He held a printed report in his hand, the cover of which read "Predictive Closed-Loop Algorithm V1.0 - Test Report".
Yang Ping sat behind his desk, with several documents spread out in front of him and a cup of tea that had gone cold.
"sit!"
Tang Shun didn't sit down. He stood in front of the whiteboard and pasted the report materials page by page onto it, securing them with magnetic pins. This was a habit he learned from Yang Ping: reporting wasn't about reading from a script, but about drawing diagrams, turning logic into a visual structure.
He said, "The core idea is to use an LSTM network for time series prediction. The input is the electrical activity characteristics of M7 over the past ten minutes, and the output is the discharge probability distribution for the next five minutes. When the predicted abnormal discharge probability exceeds a threshold, light stimulation is initiated thirty seconds in advance, instead of waiting for the abnormality to occur before taking remedial action."
"How was the 30-second lead time determined?"
"In our simulation tests, we ran 3,000 sets of data and found that 30 seconds is the optimal balance point. If the lead time is too short, there's not enough time to intervene; if the lead time is too long, the false alarm rate increases. At 30 seconds, the product of sensitivity and specificity is at its maximum."
What is the false alarm rate?
"8.3% means that one in every twelve predictive triggers is a false positive. But the cost of a false positive is minimal; it's just one extra light stimulus, with no side effects."
Yang Ping nodded, signaling him to continue. Tang Shun was indeed an outstanding multi-talented individual; he could even write algorithms himself.
Tang Shun turned to the next page: "In terms of hardware optimization, I compressed the data dimension from the original 256 dimensions to 64 dimensions, using principal component analysis combined with t-SNE dimensionality reduction. The accuracy loss mainly comes from the compression process, but we compensated at the model level by using a deeper network structure to extract features."
"Is there enough training data?"
“It’s not enough,” Tang Shun said honestly. “The M7 data only has five weeks of data, and the sample size is very small. I used transfer learning, first pre-training on a public dataset of rats, and then fine-tuning it using the M7 data.”
Yang Ping frowned: "The electrical activity patterns of rats and primates differ greatly. Can the effectiveness of transfer learning be guaranteed?"
Tang Shun explained frankly, "I can't guarantee that, so I added an adaptive module to the model that can dynamically adjust the weights based on M7's real-time feedback. If the features learned through transfer learning become unsuitable, the model will gradually de-emphasize them and instead rely on M7's own data."
Yang Ping remained silent for a moment. Then he stood up, walked to the whiteboard, and stared at the complex network structure diagram.
"Tang Shun, do you know where the flaws in this plan lie?"
Tang Shun paused for a moment. “The drawback…” He thought for a moment, “The adaptive module needs to learn online, which means the model will constantly change. The changes themselves can introduce instability; if M7’s state suddenly changes, such as due to infection, stress, or a drug response, the model might not have enough time to adapt.”
"Yes, anything else?"
“And also…” Tang Shun’s brows furrowed even more deeply, “The internal state of an LSTM network is a black box; we can’t see why it makes a certain prediction. If it’s wrong, we don’t know where the error is, and therefore we can’t manually correct it.”
“Yes,” Yang Ping nodded again, “there’s one more crucial one.”
Tang Shun looked at him, waiting for an answer.
Yang Ping said, “You assume that M7’s future is predictable, but life isn’t a machine. M7’s mood today, its sleep last night, and even what it dreamed about can all affect its electrical activity today. You can’t quantify these variables, nor can you input them into a model. Your prediction is essentially based on the assumption that ‘yesterday was like the day before yesterday, so today will be like yesterday.’ But this assumption is always fragile in biological systems.”
Tang Shun fell silent, staring at the network diagram on the whiteboard. Suddenly, the lines and nodes seemed unfamiliar. The intricate structure he had built over forty-eight hours revealed cracks in its foundation in Yang Ping's words.
“Then what should we do?” he asked.
“It’s not that we shouldn’t use it,” Yang Ping said. “When we do use it, we need to know its limitations. Treat it as an aid, not as absolute truth. The final decision still rests with humans. The model tells you ‘it might be abnormal,’ and you determine ‘whether it really is abnormal.’”
He walked back to his desk, took a document from the drawer, and handed it to Tang Shun.
"This is the clinical observation guideline I wrote. Once M7 enters a stable phase, we will begin preparing for human trials. One clause in the guideline states that all model-based predictions must be independently confirmed by at least two physicians before implementation."
Tang Shun took the document and quickly skimmed through it. His gaze lingered on a particular page, where a line was marked in red: "Technology cannot replace judgment, and data cannot replace experience."
“I understand,” he said.
In the evening, Elena was preparing dinner in the research institute's restaurant kitchen.
The institute’s cafeteria has a self-service kitchen. Elena spread a checkered tablecloth she brought from Germany on the counter, and then took out ingredients from an insulated bag—German sausages, sauerkraut, mashed potatoes, and a pot of hot apple pie.
“Ms. Elena,” Lina suddenly appeared, peeking in, “do you need any help?”
“Call them over for me,” Elena said. “Just tell them dinner is ready.”
Lina turned to call for someone. She went to the lab first, where Hans and Eva were discussing data. Upon hearing the words "Dinner's ready," Hans's stomach immediately rumbled loudly. Eva glanced at him, her lips twitching slightly—her way of expressing a smile.
Then Lina went to the animal room, where Fritz was sitting next to M7's cage, holding the copy of "Journey to the West" and reading the chapter about Sun Wukong's three battles with the White Bone Demon.
"Fritz, dinner's ready."
"I'm watching over M7; it needs someone to keep it company tonight."
"Ms. Elena said that M7 also needs you to eat so that you have the energy to keep it company."
Fritz paused for a moment. He closed his book, stood up, and followed Lena to the dining room. Reaching the door, he glanced back at M7. M7 was lying in its cage, eyes half-open, hind legs slightly curled. Its breathing was steady, as if it were having a sweet dream.
“I’ll be right back, brother!” Fritz closed the door behind him.
A small room in the research institute's cafeteria was packed with people.
Yang Ping sat in a corner chair with a plate of sausages and mashed potatoes in front of him. He didn't touch his knife and fork, just watched the others. Weber and Elena sat together, sharing a piece of apple pie, occasionally exchanging glances without speaking. Hans and Lina were arguing about the dialectal German word for "sausage." Hans said Bavarians call it "Wurst," while Lina said Berliners call it "Kiezwurst." They argued heatedly, but not a single bite was left on their plates. Eva sat alone on the other side, eating slowly, chewing each bite for a long time, as if savoring some complex flavor.
Tang Shun was the last to arrive. He carried a plate and sat down next to Yang Ping.
"I've reconsidered the algorithm."
"Hmm," Yang Ping responded without looking up.
"You're right, the model has boundaries. I plan to add an uncertainty quantification module, so that the model outputs a confidence level while making predictions. When the confidence level is low, it will automatically prompt for manual intervention."
Yang Ping looked up and said, "Okay, but don't make it too complicated. Simple things stand the test of time."
"clear."
Tang Shun started eating, eating very quickly, as if he wanted to make up for all the meals he had missed in the past forty-eight hours. Yang Ping glanced at him, said nothing, and simply pushed the untouched sausage from his plate towards him.
"Aren't you going to eat?" Tang Shun asked.
“I’m not hungry,” Yang Ping said. “You eat.”
Tang Shun didn't stand on ceremony. He forked a sausage, took a big bite, and juice splattered onto his chin. He wiped it with the back of his hand and continued eating.
Weber watched this scene unfold. He recalled his team dinners in Germany, where everyone dressed formally, seated at either end of a long table according to rank. Conversations revolved around grant applications, paper citations, and academic politics. No one would push food from their plate to another, because that implied "I don't need this," a subtle form of offense in the German context.
But here, Yang Ping pushes the sausage to Tang Shun without any deeper meaning; it's as simple as breathing.
“Professor Weber,” Elena nudged his arm, “eat it, it won’t taste good cold.”
Weber lowered his head and forked a piece of mashed potato. The mashed potato was soft, with the aroma of butter and milk, reminding him of many years ago when Elena first cooked for him. At that time, they were still in Heidelberg, living in an unheated attic, relying on an electric heater for warmth in the winter. Elena used a small electric stove to cook a pot of mashed potatoes, which the two of them shared, eating until they were covered in sweat.
“Karl,” Elena said softly, “what are you thinking about?”
Weber said, "Think back to the past, think back to when we were young."
"Isn't being young a good thing?"
"Okay, but it's even better now."
Elena smiled. She didn't ask "why," because she knew the answer. When he was young, Weber had a fire in his heart, but that fire burned others and himself. Now, the fire was still there, but the direction it burned had changed. It was no longer about proving himself, but about doing something well. Even after he won the Nobel Prize, he still felt he hadn't done it well, but now he felt he could do it well.
After dinner, Yang Ping went to the rooftop alone.
The rooftop of the Sanbo Institute was an abandoned terrace, usually locked, but Yang Ping had a key. He opened the door, walked to the railing, and looked at the city lights in the distance. He often liked to look out like this; it broadened his horizons.
He heard footsteps behind him, but without turning around, he knew who it was.
“Professor Weber,” he said, “you should go to sleep.”
“I can’t sleep,” Weber said, walking over to him and leaning on the railing. “I’m still jet-lagged.”
"You've been here for three months, are you still experiencing jet lag?"
“It’s not the German time difference,” Weber said, “it’s an age-related time difference. As people get older, their sleep is like an old battery that can’t be fully recharged.”
Yang Ping didn't reply. The two stood side by side, looking at the lights in the distance. The night breeze was a bit chilly; Weber wrapped his coat tighter around himself, but Yang Ping didn't—perhaps an advantage of being younger.
“Professor Yang, sometimes I really don’t understand why you always have so many novel ideas, and why you can always turn those ideas into a practical path. Sometimes it seems chaotic at the beginning, and it feels like nothing can be accomplished, but as you work on it, the path gradually becomes clear, and you can gradually see a clear future. For example, the three-dimensional guided gene theory, and later K therapy. When you introduced it, many people, including me, simply did not believe it, because I thought it was too absurd, impossible, and simply a fog, with no possibility in sight. But slowly, everything became clear.”
Weber continued.
"What I mean is, when things start out like a mess, why do you dare to begin, why are you able to navigate through the fog? Intuition?
His question is a bit difficult to answer, but I still have to answer it.
Yang Ping thought for a moment and said, "Well, you could call it intuition, but it's not really. Intuition is actually the sedimentation of knowledge accumulated to a certain level. For example, standing here now, I know that's south, but you don't. Even if I'm anywhere in this city, surrounded by skyscrapers, I can immediately tell which way is north, south, east, and west, but you can't. But if you go to Berlin, you can do it, but I can't. Because we've lived in our respective cities for so many years, we're very familiar with them. This familiarity has reached the point where analysis is unnecessary; it's a sedimentation of familiarity. It's the same in scientific research. Is it intuition? Actually, no, it's a sedimentation in the brain after a certain level of accumulation."
Weber nodded: "I think I understand. In fact, the so-called intuition in scientific research does not exist. It is just a kind of correct accumulation in the brain, just like a smart and experienced policeman who can always easily identify thieves in a crowd."